Do not restore pc when returning from top most fiber frame.
Also add JANET_DEBUG config define for various debugging related
configurations. In fiber.c, when debug is enabled we reallocate the
entire stack everytime we push a frame to help uncover use after free
errors.
Lazy verification makes it easier to not leave funcenvs
in an invalid state, as well as be more precise with the validation.
We needed to verify the FuncEnvs actually pointed to a stack frame if
they were of the "on-stack" variant. There was some minor checking
before, but it was not enough to prevent func envs from pointing to
memory that was off of the fiber stack, overlapping stack frames, etc.
Using a bitset to indicate which stack values are upvalues, we
can more accurately track when a reference to a stack value
persists after the stack frame exits.
If possible, this will reduce the need to marshal fibers
in many cases. Also add this logic to the GC so holding a closure
that originally came from a fiber that crashed does not cause that fiber
to hang around forever.
These additions, along with the change that user signals 0-4 cannot
be resumed, allow delimited continuation semantics, while repsecting
existing forms like `defer`, `with`, `with-vars`, etc.
Required a few changes to APIs, namely janet_root_fiber()
to get topmost fiber that is active in the current scheduler.
This is distinct from janet_current_fiber(), which gets the bottom
most fiber in the fiber stack - it might have a parent, and so cannot
be reliably resumed.
This is the kind of situation that makes symmetric coroutines more
attractive.
This way we can support fewer build configurations. Also, remove
all undefined behavior due to use of memcpy with NULL pointers. GCC
was exploiting this to remove NULL checks in some builds.
While C functions are not re-entrant, signaling from a C function
can be used to implement async returns. When resuming a fiber that
signalled from within a C function, the fiber is started after the
instruction that emitted the signal. The resume argument is used
as the return result from the c function.
Also make integer to size_t casts explicit rather than relying on
int32_t * sizeof(x) = size_t. This is kind of a personal preference for
this problem.
Because we use an amalgated build, feature
test macros should be set in a single file that
is included before any other headers, and is placed
at the top of the amalgamated build.
We normally only track memory allocated with janet_gcalloc, but
if only a few very large normal memory blocks are allocated, the GC
will never run. Se simply need to increment a count when we allocate
memory so that the next time we enter the VM, we will be able to
run a collection if needed.
This makes it easier to document functions that
take keyword arguments and also prevents some allocations
with these functions. Before, this was possible via normal
variadic functions but created an intermediate tuple, and
the generated docstrings did not document the keys.
This should help address #81. Also hide janet_exit
and janet_assert, as they are really meant for internal usage.
I have not verified that this yet actually works with Rust's
bindgen.
- Allow passing a table to fibers, which make fiber level scope easier.
- Add fiber/getenv, fiber/setenv, dyn, and setdyn
- Remove meta, *env*, and *doc-width*
- Some functions changed dignatures, and no longer take an env
A consistent style should help with contributors and
readability. We use astyle as the formatter as can make a pretty
good approximation of the current style and my preferred style.
Astyle can be found at http://astyle.sourceforge.net/astyle.html
This should speed up start time and reduce malloc/free
usage to about 15% of what is what previously for startup.
The current cost is slightly larger binary as the representaion
of the image is currently less compact than source code.
to make janet function calls easier and faster from C (still
needs an object pool for fibers, though). Fix bug in scan-number
and add many more peg tests.
Error handling is implemented with setjmp/longjmp so code
can be more concise. This required a very large but straight forward refactor for all
of the libraries.
