Strict xsloader #1
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
atoomic
reviewed
Oct 7, 2020
toddr
force-pushed
the
xs_dynaloader
branch
12 times, most recently
from
7812c11 to
59724ad
Compare
Most implementations do not have a problem with two getenv()'s running simultaneously in different threads. But Posix doesn't require such good behavior. This adds a simple probe to test the current system.
get_env() needs to lock other threads from writing to the environment while it is executing. It may need to have an exclusive lock if those threads can clobber its buffer before it gets a chance to save them. The previous commit has added a Configure probe which tells us if that is the case. This commit uses it to select which type of mutex to use.
This fixes GH Perl#18341 There are problems with getenv() on threaded perls wchich can lead to incorrect results when compiled with PERL_MEM_LOG. Commit 0b83dfe fixed this for some platforms, but as Tony Cook, pointed out there may be standards-compliant platforms that that didn't fix. The detailed comments outline the issues and (complicated) full solution.
# Conflicts: # pod/perldelta.pod
This rearranges some paragraphs that really belong together, and clarifies the result is always in English
And give locale's alternative
In 83141f1, the maximum line length of all lines added was 77, which should have passed the 78 maximum used in t/porting/podcheck.t. There were no tabs in the allegedly offending lines and no '=over' directives in operation. So it's unclear why this failed. Nonetheless, as additional commits are pending, it's expedienct for us to rebreak the lines manually to get the test program to pass and figure out the problem another day.
Changes in inline comments only; responds to concern raised in commit message for 363d555.
The Time::HiRes build was broken by 1d96b9c and requires a fix similar to daacfc6 for mingw runtimes < 5.0. A definition of timespec was added in mingwrt-5.0. The compilers from mingw-w64 (which can be singled out by their definition of __MINGW64_VERSION_MAJOR) were not affected since they all have timespec defined anyway. For me, this fixes the build with my mingw.org compilers from 3.4.5 to 4.8.1 inclusive. These are using runtimes 3.15.2 to 4.0.3 (which actually reports itself as 3.20.0!) inclusive. Note that builds with mingw.org compilers using mingwrt >= 5.0 are still broken due to a different problem involving mkstemp(), as documented in 8a217c9. See GH#15446 for more details.
MinGW runtime version 3.21 added a definition of mkstemp(), so requires a fix similar to f33b2f5 for MinGW-w64's runtime 4.0 onwards. Based on a patch by Dan Collins on GH#15446. This commit also tweaks 43b3b04, having discovered that mingw runtime 3.22 also contains a definition of timespec. For me, this fixes the build with my mingw.org 5.3.0 compiler using any of the mingw runtimes 3.21, 3.22.4 or 5.0 without breaking older versions such as 4.9.3 with the 3.20 runtime (or even with the withdrawn 4.0.3 runtime, which had __MINGW32_MAJOR/MINOR_VERSION set to 3.20 whilst adding new a __MINGW_MAJOR/MINOR_VERSION set to 4.0). However, 6.3.0 and 7.3.0 have other issues when compiling win32sck.c, while 8.2.0 and 9.2.0 have other issues again when compiling win32.c. See GH#18510 for more details. Also, C++ mode builds with some MinGW/MinGW-w64 compilers are still broken, as documented in 8a217c9. See GH#16459 for more details.
Makefile.mk is redundant with GNUmakefile. See https://www.nntp.perl.org/group/perl.perl5.porters/2021/01/msg258848.html for more details. We planned to remove it shortly after the introduction of GNUmakefile but that slipped through the cracks for some reason: Perl#14341
Previously the default libpth consisted of just a single folder and
failed to include the directory that contains the majority of the
libraries.
This is a fairly important issue but no one noticed it for two reasons:
1. EU::MM on Windows *always* links XS modules with the libraries from
$Config{libs}, so you'd notice that linking doesn't work only if you
needed a library that isn't listed there.
2. Strawberry Perl has a workaround for this issue[1].
I'm only using MinGW-w64 compilers, so I have no idea how library paths
work on MinGW.org builds. It's possible that the previous libpth worked
fine with them. Either way, this commit only adds new paths to libpth, it
doesn't modify the one that was already there, so it's unlikely it will
break anything.
[1] - https://github.com/StrawberryPerl/Perl-Dist-Strawberry/blob/2112b8a590882e913e98e4aa2dced4f34c4fea79/lib/Perl/Dist/Strawberry/Step/InstallPerlCore.pm#L136
PerlIO::encoding has a $fallback variable that allows one to set the behavior on a encoding/decoding error, for example to make it throw an exception on error. What is not documented (actually the example in the documentation is even missing this) is that PerlIO::encoding needs the (equally undocumented) Encode::STOP_AT_PARTIAL flag to be set, otherwise a multi-byte character spanning buffer boundaries will be interpreted as two invalid byte sequences. I could have fixed the documentation, but instead I fixed the code to always pass this flag to Encode, simplifying the use and making the current documentation correct again.
Encode allows one to pass a coderef instead of a set of flags to handle. This however doesn't allow one to pass STOP_AT_PARTIAL, which means it has always been buggy on buffer boundaries. With my new automatic STOP_AT_PARTIAL passing this would result in an unpredictable value. Instead we now disallow it in PerlIO::encoding.
It's also set from encoding.pm, doing it double serves no purpose
Setting $PerlIO::encoding::fallback to any value containing LEAVE_SRC will result in an infinite loop of the first buffer of input. This is never desirable.
The ticket mentions yet another message, not addressed in this commit, "Insecure private-use override". That message is part of a hook for a so-far unimplemented module, so it actually doesn't ever get raised. Committer: One correction per Grinnz comment in Perl#18491
Originally missing the word "at", but rather than just adding it, this wording seemed nicer.
CxTRYBLOCK would be confusing when we add a real CxTRY for try/catch Also renames the associated CXp_TRYBLOCK flag to CXp_EVALBLOCK
* Add feature, experimental warning, keyword * Basic parsing * Basic implementation as optree fragment See also Perl#18504
Add warnings to tests.
Now we're using our, there's no value to attempting to support versions below 5.6.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is just an idea. Currently these tests fail:
Further, you can see this error which is disturbing: