laria/apfl
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
apfl/src/eval.c

330 lines
8.1 KiB
C
Raw Normal View History

Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
#include <assert.h>
#include "apfl.h"
Introduce allocator abstraction We now no longer call malloc/free/... directly, but use an allocator object that is passed around. This was mainly done as a preparation for a garbage collector: The collector will need to know, how much memory we're using, introducing the collector abstraction will allow the GC to hook into the memory allocation and observe the memory usage. This has other potential applications: - We could now be embedded into applications that can't use the libc allocator. - There could be an allocator that limits the total amount of used memory, e.g. for sandboxing purposes. - In our tests we could use this to simulate out of memory conditions (implement an allocator that fails at the n-th allocation, increase n by one and restart the test until there are no more faked OOM conditions). The function signature of the allocator is basically exactly the same as the one Lua uses.
2022-02-08 21:53:13 +00:00
#include "alloc.h"
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
#include "bytecode.h"
#include "compile.h"
#include "context.h"
apfl_debug_print_val: Use format writer abstraction
2022-06-05 20:06:33 +00:00
#include "format.h"
Implement simple variable support We can now assign to variables and retrieve their values. Not all assignables are implemented yet, but still pretty cool :).
2022-01-06 21:53:26 +00:00
#include "hashmap.h"
Ditch that weird internal.h header
2022-04-22 21:17:28 +00:00
#include "strings.h"
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
#include "value.h"
static void
Add apfl_stackidx typedef This makes it clear that a value of that type is meant as an index into the stack.
2022-02-10 21:39:39 +00:00
stack_must_drop(apfl_ctx ctx, apfl_stackidx index)
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
assert(apfl_stack_drop(ctx, index));
Partially implement pattern matching assignments We're still missing predicates (need to be able to call functions for that one) and assignments into dictionaries. But we now can deconstruct a list and check against constants. So things like this work now: [1 foo ~bar [a b]] = [1 "Hello" 2 3 4 [5 6]] # foo is: "Hello" # bar is: [2 3 4] # a is: 5 # b is: 6 Pretty cool :)
2022-01-14 22:16:19 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static bool
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
get_argument(size_t *i, struct instruction_list *ilist, union instruction_or_arg *arg)
Partially implement pattern matching assignments We're still missing predicates (need to be able to call functions for that one) and assignments into dictionaries. But we now can deconstruct a list and check against constants. So things like this work now: [1 foo ~bar [a b]] = [1 "Hello" 2 3 4 [5 6]] # foo is: "Hello" # bar is: [2 3 4] # a is: 5 # b is: 6 Pretty cool :)
2022-01-14 22:16:19 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
if (*i >= ilist->len) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
return false;
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
}
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
*arg = ilist->instructions[(*i)++];
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
return true;
}
static void
must_get_argument(apfl_ctx ctx, size_t *i, struct instruction_list *ilist, union instruction_or_arg *arg)
{
if (!get_argument(i, ilist, arg)) {
apfl_raise_const_error(ctx, APFL_RESULT_ERR, apfl_messages.corrupted_bytecode);
}
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
variable_get(apfl_ctx ctx, struct apfl_string *name)
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
struct apfl_value value;
if (!apfl_scope_get(ctx->scope, name, &value)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
// TODO: Include variable name in error
apfl_raise_const_error(ctx, APFL_RESULT_ERR, apfl_messages.variable_doesnt_exist);
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_stack_must_push(ctx, value);
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
variable_set(apfl_ctx ctx, struct apfl_string *name, bool keep_on_stack)
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
struct apfl_value value;
if (!apfl_stack_get(ctx, &value, -1)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_invalid_stackidx(ctx);
values: Split lists into read-only lists and editable lists This is analogous to dictionaries and ensures that no circular references can be created when using the exported API in apfl.h. This also changes apfl_value_copy into apfl_value_incref to better reflect what it does and to reflect that it is no longer an operation that can fail.
2022-01-04 22:11:38 +00:00
}
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
if (!apfl_scope_set(&ctx->gc, ctx->scope, name, value)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_alloc_error(ctx);
Implement evaluating dictionary literals We can now have dictionaries with keys and values of arbitrary types. Very cool! :)
2022-01-04 20:51:44 +00:00
}
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
if (keep_on_stack) {
// If the value should be kept on the stack, the value is now in two
// places. We need to set the COW flag to prevent mutations of one copy
// affecting the other one.
value = apfl_value_set_cow_flag(value);
} else {
stack_must_drop(ctx, -1);
Implement evaluating dictionary literals We can now have dictionaries with keys and values of arbitrary types. Very cool! :)
2022-01-04 20:51:44 +00:00
}
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
variable_new(apfl_ctx ctx, struct apfl_string *name)
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
if (!apfl_scope_create_var(&ctx->gc, ctx->scope, name)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_alloc_error(ctx);
Implement evaluating dictionary literals We can now have dictionaries with keys and values of arbitrary types. Very cool! :)
2022-01-04 20:51:44 +00:00
}
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
evaluate(apfl_ctx ctx, size_t *i, struct instruction_list *ilist)
Implement evaluating container member access
2022-01-04 23:25:41 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
union instruction_or_arg arg;
Implement evaluating container member access
2022-01-04 23:25:41 +00:00
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
assert(*i < ilist->len);
switch (ilist->instructions[(*i)++].instruction) {
case INSN_NIL:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_push_nil(ctx);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_TRUE:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_push_bool(ctx, true);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_FALSE:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_push_bool(ctx, false);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_NUMBER:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
apfl_push_number(ctx, arg.number);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_STRING:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
apfl_stack_must_push(ctx, (struct apfl_value) {
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
.type = VALUE_STRING,
.string = arg.string,
});
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_LIST:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
apfl_list_create(ctx, arg.count);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_LIST_APPEND:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_list_append(ctx, -2, -1);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_LIST_EXPAND_INTO:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_list_append_list(ctx, -2, -1);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_DICT:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_dict_create(ctx);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_DICT_APPEND_KVPAIR:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_dict_set(ctx, -3, -2, -1);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_GET_MEMBER:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_get_member(ctx, -2, -1);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_VAR_NEW:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
variable_new(ctx, arg.string);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_VAR_GET:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
variable_get(ctx, arg.string);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_VAR_SET:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
variable_set(ctx, arg.string, true);
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_NEXT_LINE:
ctx->execution_line++;
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
return;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
case INSN_SET_LINE:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
must_get_argument(ctx, i, ilist, &arg);
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
ctx->execution_line = arg.count;
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
return;
Implement evaluating container member access
2022-01-04 23:25:41 +00:00
}
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
assert(false);
Implement evaluating container member access
2022-01-04 23:25:41 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
evaluate_list(apfl_ctx ctx, struct instruction_list *ilist)
Implement evaluating container member access
2022-01-04 23:25:41 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
ctx->execution_line = ilist->line;
size_t i = 0;
while (i < ilist->len) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
evaluate(ctx, &i, ilist);
Implement evaluating container member access
2022-01-04 23:25:41 +00:00
}
}
Implement evaluating dictionary literals We can now have dictionaries with keys and values of arbitrary types. Very cool! :)
2022-01-04 20:51:44 +00:00
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
eval_expr_inner(apfl_ctx ctx, struct apfl_expr expr)
Implement simple variable support We can now assign to variables and retrieve their values. Not all assignables are implemented yet, but still pretty cool :).
2022-01-06 21:53:26 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
struct instruction_list *ilist = apfl_instructions_new(&ctx->gc, expr.position.line);
if (ilist == NULL) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_alloc_error(ctx);
Partially implement pattern matching assignments We're still missing predicates (need to be able to call functions for that one) and assignments into dictionaries. But we now can deconstruct a list and check against constants. So things like this work now: [1 foo ~bar [a b]] = [1 "Hello" 2 3 4 [5 6]] # foo is: "Hello" # bar is: [2 3 4] # a is: 5 # b is: 6 Pretty cool :)
2022-01-14 22:16:19 +00:00
}
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
if (!apfl_gc_tmproot_add(&ctx->gc, GC_OBJECT_FROM(ilist, GC_TYPE_INSTRUCTIONS))) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_alloc_error(ctx);
Partially implement pattern matching assignments We're still missing predicates (need to be able to call functions for that one) and assignments into dictionaries. But we now can deconstruct a list and check against constants. So things like this work now: [1 foo ~bar [a b]] = [1 "Hello" 2 3 4 [5 6]] # foo is: "Hello" # bar is: [2 3 4] # a is: 5 # b is: 6 Pretty cool :)
2022-01-14 22:16:19 +00:00
}
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
struct apfl_error error;
if (!apfl_compile(&ctx->gc, expr, &error, ilist)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_error_object(ctx, error);
Implement simple variable support We can now assign to variables and retrieve their values. Not all assignables are implemented yet, but still pretty cool :).
2022-01-06 21:53:26 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
evaluate_list(ctx, ilist);
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
eval_expr(apfl_ctx ctx, struct apfl_expr expr)
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
{
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
size_t tmproots = apfl_gc_tmproots_begin(&ctx->gc);
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
eval_expr_inner(ctx, expr);
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
apfl_gc_tmproots_restore(&ctx->gc, tmproots);
Start work on evaluating expressions Right now we're only evaluating bool/nil/number/string/list literals, but it's a start :).
2022-01-02 16:19:54 +00:00
}
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
Use formatter for printing values
2022-04-22 21:13:01 +00:00
bool
apfl_debug_print_val: Use format writer abstraction
2022-06-05 20:06:33 +00:00
apfl_debug_print_val(apfl_ctx ctx, apfl_stackidx index, struct apfl_format_writer w)
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
{
struct apfl_value value;
Implement mark&sweep garbage collection and bytecode compilation Instead of the previous refcount base garbage collection, we're now using a basic tri-color mark&sweep collector. This is done to support cyclical value relationships in the future (functions can form cycles, all values implemented up to this point can not). The collector maintains a set of roots and a set of objects (grouped into blocks). The GC enabled objects are no longer allocated manually, but will be allocated by the GC. The GC also wraps an allocator, this way the GC knows, if we ran out of memory and will try to get out of this situation by performing a full collection cycle. The tri-color abstraction was chosen for two reasons: - We don't have to maintain a list of objects that need to be marked, we can simply grab the next grey one. - It should allow us to later implement incremental collection (right now we only do a stop-the-world collection). This also switches to a bytecode based evaluation of the code: We no longer directly evaluate the AST, but first compile it into a series of instructions, that are evaluated in a separate step. This was done in preparation for inplementing functions: We only need to turn a function body into instructions instead of evaluating the node again with each call of the function. Also, since an instruction list is implemented as a GC object, this then removes manual memory management of the function body and it's child nodes. Since the GC and the bytecode go hand in hand, this was done in one (giant) commit. As a downside, we've now lost the ability do do list matching on assignments. I've already started to work on implementing this in the new architecture, but left it out of this commit, as it's already quite a large commit :)
2022-04-11 20:24:22 +00:00
if (!apfl_stack_pop(ctx, &value, index)) {
apfl_debug_print_val: Use format writer abstraction
2022-06-05 20:06:33 +00:00
FMT_TRY(apfl_format_put_string(w, "apfl_debug_print_val: Invalid stack index "));
FMT_TRY(apfl_format_put_int(w, (int)index));
FMT_TRY(apfl_format_put_string(w, "\n"));
Use formatter for printing values
2022-04-22 21:13:01 +00:00
return true;
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
}
apfl_debug_print_val: Use format writer abstraction
2022-06-05 20:06:33 +00:00
return apfl_value_print(value, w);
Convert to Lua-style stack API Only for evaluating expressions for now and right now the only exposed operation is to debug print a value on the stack, this obviously needs to be expanded. I've done this for two reasons: 1. A Lua-style stack API is much nicer to work with than to manually manage refcounts. 2. We'll soon need a more sophisticated garbage collector (if you even want to count the refcounting as garbage collection). For this, the GC will need root objects to start tracing for live objects, the stack will be one of these roots.
2022-01-20 21:45:09 +00:00
}
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
struct apfl_iterative_runner_data {
apfl_ctx ctx;
apfl_tokenizer_ptr tokenizer;
apfl_parser_ptr parser;
enum apfl_result result;
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
bool with_error_on_stack;
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
bool end;
};
apfl_iterative_runner
apfl_iterative_runner_new(apfl_ctx ctx, struct apfl_source_reader reader)
{
apfl_iterative_runner runner = ALLOC_OBJ(ctx->gc.allocator, struct apfl_iterative_runner_data);
if (runner == NULL) {
return NULL;
}
apfl_tokenizer_ptr tokenizer = apfl_tokenizer_new(ctx->gc.allocator, reader);
if (tokenizer == NULL) {
FREE_OBJ(ctx->gc.allocator, runner);
return NULL;
}
apfl_parser_ptr parser = apfl_parser_new(ctx->gc.allocator, apfl_tokenizer_as_token_source(tokenizer));
if (parser == NULL) {
FREE_OBJ(ctx->gc.allocator, runner);
apfl_tokenizer_destroy(tokenizer);
return NULL;
}
*runner = (struct apfl_iterative_runner_data) {
.ctx = ctx,
.tokenizer = tokenizer,
.parser = parser,
.result = APFL_RESULT_OK,
.end = false,
};
return runner;
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
handle_parse_error(apfl_ctx ctx, struct apfl_error error)
{
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
enum apfl_result errtype = APFL_RESULT_ERR;
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
if (apfl_error_is_fatal_type(error.type)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
errtype = APFL_RESULT_ERR_FATAL;
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
}
Push some errors as const strings onto the stack
2022-04-24 14:10:35 +00:00
const char *const_str = apfl_error_as_const_string(error);
if (const_str != NULL) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_const_error(ctx, errtype, const_str);
Push some errors as const strings onto the stack
2022-04-24 14:10:35 +00:00
}
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
struct apfl_string string;
if (!apfl_error_as_string(error, ctx->gc.allocator, &string)) {
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_alloc_error(ctx);
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
if (!apfl_move_string_onto_stack(ctx, string)) {
apfl_raise_alloc_error(ctx);
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_raise_error_with_type(ctx, -1, errtype);
Push formatted parser error on stack during evaluation This way we can see the parse errors again in evaluation mode Not fully fleshed out yet: We simply use apfl_debug_print_val to dump the top of the stack (the formatted error) to stderr but don't nicely handle if there is nothing on the stack (apfl_debug_print_val will print a rather cryptic "stack index invalid" error). Also the whole dance we need to do to put the formatted error onto the stack feels rather awkward (there should probably a function for this) and we also should probably try to push an error description on the stack in case this moving-string-to-stack business fails. Now "only" all other errors need to be put on the stack as a string :)
2022-04-21 20:40:27 +00:00
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
static void
iterative_runner_next_protected(apfl_ctx ctx, void *opaque)
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
{
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
apfl_iterative_runner runner = opaque;
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
switch (apfl_parser_next(runner->parser)) {
case APFL_PARSE_OK:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
eval_expr(ctx, apfl_parser_get_expr(runner->parser));
return;
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
case APFL_PARSE_ERROR:
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
handle_parse_error(runner->ctx, apfl_parser_get_error(runner->parser));
return;
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
case APFL_PARSE_EOF:
runner->end = true;
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
return;
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
}
assert(false);
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
}
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
bool
apfl_iterative_runner_next(apfl_iterative_runner runner)
{
if (runner->end) {
return false;
}
apfl_stack_clear(runner->ctx);
runner->with_error_on_stack = false;
runner->result = apfl_protected(
runner->ctx,
iterative_runner_next_protected,
runner,
&runner->with_error_on_stack
);
return !runner->end;
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
}
enum apfl_result
apfl_iterative_runner_get_result(apfl_iterative_runner runner)
{
return runner->result;
}
Implement exceptions-like error handling Most functzions will no longer return an enum apfl_result, but will raise an error that bubbles up.
2022-06-24 21:13:44 +00:00
bool
apfl_iterative_runner_has_error_on_stack(apfl_iterative_runner runner)
{
return runner->with_error_on_stack;
}
No longer evaluate exprs directly
2022-04-21 19:15:20 +00:00
void
apfl_iterative_runner_destroy(apfl_iterative_runner runner)
{
if (runner == NULL) {
return;
}
apfl_parser_destroy(runner->parser);
apfl_tokenizer_destroy(runner->tokenizer);
FREE_OBJ(runner->ctx->gc.allocator, runner);
}
Reference in a new issue Copy permalink