Problem: Make CI checks more strict
Solution: Add -Wstrict-prototypes -Wmissing-prototypes to CI,
fix uncovered problems
Add -Wstrict-prototypes -Wmissing-prototypes warnings check to CI
Add two new warnings to CI, silence some Perl related build-warnings:
- `strict-prototypes` helps prevent declaring a function with an empty
argument list, e.g. `int func()`. In C++, that's equivalent to `int
func(void)`, but in C, that means a function that can take any number
of arguments which is rarely what we want.
- `missing-prototypes` makes sure we use `static` for file-only internal
functions. Non-static functions should have been declared on a
prototype file.
- Add `no-compound-token-split-by-macro` to the perl cflags, since it
throws out a bunch of perl-related warnings that make the CI log
unnecessary verbose and hard to read. This seems to happen only with
clang 12 and above.
When applying those changes, it already uncovered a few warnings, so fix
up the code as well (fix prototypes, make the code static, remove
shadowed var declaration)
GTK header needs to have #pragma warning suppressiong because GTK2
headers will warn on `-Wstrict-prototypes`, and it's included by gui.h
and so we can't just turn off the warning in a couple files.
closes: #13223closes: #13226
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Yee Cheng Chin <ychin.git@gmail.com>
Problem: Perl interface has problems with load PL_current_context
Solution: Fix Perl interface to load PL_current_context from library
In #12914, in order to fix an issue with Perl 5.36 dynamic builds, (that
version introduced a thread-local `PL_current_context`), the file added
the variable manually so we can satisfy the linker. However, the
variable is a different one from the one in the library, so there could
be unpredictable behavior. Instead, just use `dlsym` to load the context
from the library. The fact that it's thread-local doesn't matter too
much to us because Vim's interface is single-threaded so it will work
properly.
closes: #12996
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Yee Cheng Chin <ychin.git@gmail.com>
Problem: dynamically linking perl is broken
Solution: Fix all issues
This is a combination of several commits:
1) Fix if_perl.xs not being able to build on all versions of Perl (5.30)
This fixes the dynamic builds of Perl interface. The Perl interface file
previously had to manually copy and paste misc inline functions verbatim
from the Perl headers, because we defined `PERL_NO_INLINE_FUNCTIONS`
which prevents us form getting some function definitions. The original
reason we defined it was because those inline functions would reference
Perl functions that would cause linkage errors.
This is a little fragile as every time a new version of Perl comes out,
we inevitably have to copy over new versions of inline functions to our
file, and it's also easy to miss updates to existing functions.
Instead, remove the `PERL_NO_INLINE_FUNCTIONS` define, remove the manual
copy-pasted inline functions. Simply add stub implementations of the
missing linked functions like `Perl_sv_free2` and forward them to the
DLL version of the function at runtime. There are only a few functions
that need this treatment, and it's a simple stub so there is very low
upkeep compared to copying whole implementations to the file.
Also, fix the configure script so that if we are using dynamic linkage,
we don't pass `-lperl` to the build flags, to avoid accidental external
linkage while using dynamic builds. This is similar to how Python
integration works.
2) Fix GIMME_V deprecation warnings in Perl 5.38
Just use GIMME_V, and only use GIMME when using 5.30 to avoid needing to
link Perl_block_gimme. We could provide a stub like the other linked
functions like Perl_sv_free2, but simply using GIMME is the simplest and
it has always worked before.
3) Fix Perl 5.38 issues
Fix two issues:
3.1. Perl 5.38 links against more functions in their inline headers, so we
need to stub them too.
3.2. Perl 5.38 made Perl_get_context an inline function, but *only* for
non-Windows build. Fix that. Note that this was happening in Vim
currently, as it would build, but fail to run Perl code at runtime.
4) Fix Perl 5.36/5.38 when thread local is used
Perl 5.36 introduced using `_Thread_local` for the current context,
which causes inline functions to fail. Create a stub
`PL_current_context` thread local variable to satisfy the linker for
inlined functions. Note that this is going to result in a different
`PL_current_context` being used than the one used in the library, but so
far from testing it seems to work.
5) Add docs for how to build Perl for dynamic linking to work
closes: #12827closes: #12914
Signed-off-by: Christian Brabandt <cb@256bit.org>
Co-authored-by: Yee Cheng Chin <ychin.git@gmail.com>
Problem: Evaluating "expr" options has more overhead than needed.
Solution: Use call_simple_func() for 'foldtext', 'includeexpr', 'printexpr',
"expr" of 'spellsuggest', 'diffexpr', 'patchexpr', 'balloonexpr',
'formatexpr', 'indentexpr' and 'charconvert'.
Problem: Vim9: leaking memory when using continuation line.
Solution: Keep a pointer to the continuation line in evalarg_T. Centralize
checking for a next command.
Problem: Macros for MS-Windows are inconsistent, using "32", "3264 and
others.
Solution: Use MSWIN for all MS-Windows builds. Use FEAT_GUI_MSWIN for the
GUI build. (Hirohito Higashi, closes#3932)
Problem: Can execute shell commands in rvim through interfaces.
Solution: Disable using interfaces in restricted mode. Allow for writing
file with writefile(), histadd() and a few others.
Problem: Giving error messages is not flexible.
Solution: Add semsg(). Change argument from "char_u *" to "char *", also
for msg() and get rid of most MSG macros. (Ozaki Kiichi, closes
#3302) Also make emsg() accept a "char *" argument. Get rid of
an enormous number of type casts.
Problem: FEAT_WINDOWS adds a lot of #ifdefs while it is nearly always
enabled and only adds 7% to the binary size of the tiny build.
Solution: Graduate FEAT_WINDOWS.
Problem: May get ml_get error when :perldo deletes lines or switches to
another buffer. (Nikolai Pavlov, issue #1421)
Solution: Check the buffer and line every time.
