Maps directly to flags for open syscall. Ignored on Windows. We now
let the user decide whether to open fifos in nonblocking mode or not.
Manual has been extended to document the new extensions and slightly
rearranged to put the important information first.
Previous implementation was messy and contained several race
conditions (multiple open/close operations on the same file, first
checking whether the file exists before opening it). We now always use
a single open call and then optionally do an fdopen later
on (exception being :rename which contains an unavoidable race
condition between checking whether the file exists and then renaming
it later on). Also improve error messages.
We need to defer initialization forms until all dependent creation
forms have been compiled (see CLHS make-load-form). Closes#562.
Co-authored-by: Marius Gerbershagen <marius.gerbershagen@gmail.com>
When a literal appears in the file multiple times its identity should be
preserved.
CLHS 3.2.4.4:
> If two literal objects appearing in the source code for a single
> file processed with the file compiler are the identical, the
> corresponding objects in the compiled code must also be the identical.
Previously, every bytecode object created during ext::bc-compile-file
had its own vector of constants making it impossible to satisfy this
constraint. Thus, we change ext::bc-compile-file to use the same
constants vector for all bytecode objects from the same file. The
simplest way to achieve this is to use the same compiler environment
for all of the compilation process and push the read-compile loop
into the si_bc_compile_from_stream function implemented in C.
To achieve that we store the symbol name in a separate ecl_symbol slot called
cname. It is initialized from the translation slot in the structure
cl_symbol_initializer. That makes dpp and runtime always resolve function
names to the same symbol. Fixes#534.
The file symbols_list2.h was automatically generated from the file
symbols_list.h for dpp. This commit unifies said structure definition and
removes a step for generating symbols_list2.h.
From now on at runtime the structurecl_symbol_initializer has an additional
element storing the function name. That will help solving #543.
Always use the byte sized input/output operations ReadConsoleA/WriteConsoleA
and do all the conversion to unicode by ourself.
Moreover, expand the set of known encodings for Windows codepages and print
a warning if we encounter an unsupported codepage.
Fixes#582.
We can't use ecl_process_env_unsafe() == NULL to check if ECL has
booted because the return value of ecl_process_env_unsafe is
unpredictable before ECL has booted. The reason is that
ecl_process_env_unsafe calls pthread_getspecific with an uninitialized
key stored in cl_env_key. But another call to pthread_setspecific
might have already registered a key which happens to be the same as
the not yet initialized cl_env_key, yielding a non-NULL value.
The ifdef in config-internal.h was buggy since RLIMIT_STACK is only defined after
sys/resource.h has been included which it wasn't. This lead to the stack size always being
increased to at least the default of 1 MB. To fix this, we move the check for RLIMIT_STACK
to the configure script.
Improves compilation speed for single functions by about 40-50
percent. Precompiled headers are specific to the compiler version and
options in use. Due to this, we regenerate the header whenever the
compiler configuration changes.
Slot definitions are no longer a signature, but they are still needed
to update obsolete instances. Reader function name is also changed to
SI:INSTANCE-SLOTDS. SI:INSTANCE-SIG-SET name does not change, because
it sets both SLOTDS and the STAMP.
We should call make-instances-obsolete from finalize-inheritance if we
want to be conforming, because user may have added their own auxiliary
methods.
This change while being last in a serie of commits was locally the
first change which solved problems. It will enable us to implement the
fast generic dispatch after the release.
It helps to determine whether the instance needs to be updated for a
class after redefinition. Currently it is done by comparing the
signature with the instance's class slots.
Function is rewritten in C in compiler.d to remove a dependency
between the bytecodes compiler and the clos module. It may be more
performant thanks to more precise type handing, however we use a list
instead of a hashtable, so it may be slower with lookup. To assess
that we should run some benchmarks against real code -- rewriting C
code to work with a hash table should be trivial.
clos::need-to-make-load-form-p is now si::need-to-make-load-form-p and
may be called from C code as si_need_to_make_load_form_p.
The C standard allows compilers to ignore side effects of floating
point operations unless the STDC FENV_ACCESS pragma is in effect. This
could lead to spurious floating point exceptions being signaled
because the C compiler optimized a calculation that would ordinarily
not set fpe bits to one that did (observed with clang for a cast
double to cl_fixnum).
To solve this, we resurrect the ECL_MATHERR macro which was removed in
commit cb03494a6d. We handle fpe bits
only around calculations which are known to be "unsafe" in the sense
that they can produce floating point exceptions. Before the
calculation, all bits are unset and afterwards, we test for
exceptions.
The disadvantage of this method is that optimizations by the C
compiler resulting in unboxed float arithmetic might lead to
missing exceptions, because our native C compiler does not insert
ECL_MATHERR statements around these calculations.
Make sure that apart from turning on/off signaling for exceptions, we
also clear all exception bits. After the recent changes to how ECL
handles floating point exceptions, not doing that could lead to
spurious exceptions being signaled even when using the
ECL_WITH_LISP_FPE macro because an earlier calculation yielded
infinity/NaN.
Using isnan and isfinite to check for floating point exceptions
doesn't work when we want floats to be able to have infinity or NaN as
values, thus this option was removed with commit
5f71f728a3. However, we can still use
fetestexcept to explicitly check if floating point exceptions occurred
even when we can't use the feenableexcept/SIGFPE signal delivery
mechanism.
Previously, we had something like this in the
ECL_MATHERR_TEST/ECL_MATHERR_CLEAR macros, but this was not used
consistently in our codebase (the ECL_MATHERR_TEST macro was missing
in many places). Instead of error-prone testing at every point of
computation, we call fetestexcept in DO_DETECT_FPE when creating a new
float/complex float. In order to avoid having to do this twice, the
DO_DETECT_FPE2 macro is introduced.
A minor disadvantage of this strategy is that floating point
exceptions may be signaled later than they occurred.
The gmp manual (https://gmplib.org/manual/Custom-Allocation.html) states
> GMP may use allocated blocks to hold pointers to other allocated
blocks. This will limit the assumptions a conservative garbage
collection scheme can make.
Thus we can't use ecl_alloc_atomic. We could just use ecl_alloc
instead, however our implementation is already structured in such a
way that it is sufficient to use ecl_alloc_uncollectable. The reason
for that is that currently all bignums except those in the bignum
registers in the thread local environment are allocated as compact
objects so that we only need to call mpz_clear for the few non-compact
objects in the bignum registers.
See also commits 7b536161 and bd48b859.
Fixes#485.
Previously: unconditionally use the size from ECL_OPT_C_STACK_SIZE if
it is larger than the OS provided value and if possible resize the
stack at runtime.
Now: Use ECL_OPT_C_STACK_SIZE if provided. Otherwise, use a) the OS
provided value if we can resize the stack at runtime (for Unix) or b)
the new build option ECL_DEFAULT_C_STACK_SIZE (1 MB) if we can set the
stack size at link time (for Windows).
This is more robust against changes in the abi support in libffi.
We also remove unused struct and enum type definitions (left over from
our own dynamic ffi implementation?).
ecl_cs_set_org and cs_set_size had several problems:
-the size of the stack was estimated to be 1/2 of the actual value,
which could lead to wrong stack overflow errors. In particular, we
cannot assume that the stack pointer will always start at the stack
origin and increase/decrease linearly (counterexamples could
include callbacks from other threads).
-despite its name, cs_set_size did not actually set the stack size
even on systems where this is possible at runtime.
-there were several magic numbers used.
We also use the *in-all-symbols-functions* variable only for
determining which functions are to be exported during the build of ECL
itself. Otherwise, instead of specifying manually, which Lisp
functions are exported and in the core, we use the information from
symbols_list.h (i.e. we let all_symbols.d initialize all core
functions).
Previously, we assumed that the fixed and variadic arguments of a
variadic function were passed to the function in the same way. The
arm64 calling convention used by iOS breaks this assumption by passing
fixed arguments in registers or on the stack, depending on the
position, while variadic arguments are always passed on the stack.
Solving this problem while still allowing function redefinition at
runtime requires introducing additional dispatch functions. These
dispatch functions take no fixed arguments and pass all their
arguments to the actual function. This dispatch is enabled by passing
-DECL_C_COMPATIBLE_VARIADIC_DISPATCH to the C compiler.
This problem was originally identified and a solution provided by
thewhimer@gmail.com. This commit based on his work with minor
improvements.
Symbols must be associated with functions following an uniform naming
scheme because si:mangle-name does not return pre-existing string.
See #534 for details. This is a temporary solution.
