mpdecimal: Add version 4.0.0 #25896
mpdecimal: Add version 4.0.0 #25896
Conversation
Ahajha
force-pushed
the
ahajha/mpdecimal-4.0.0
branch
from
741f3b0 to
1e6dfb3
Compare
Ahajha
force-pushed
the
ahajha/mpdecimal-4.0.0
branch
from
1e6dfb3 to
9c9bb38
Compare
Ahajha
changed the title
| copy(self, "mpdecimal.h", src=mpdecdir, dst=pkg_dir / "include") | ||
| if self.options.cxx: | ||
| copy(self, "decimal.hh", src=mpdecppdir, dst=pkg_dir / "include") | ||
| copy(self, "decimal.hh", src=self.source_path / "libmpdec++", dst=pkg_dir / "include") |
There was a problem hiding this comment.
This section is weird. For some reason, in the cache folder, most of the the files are symlinks in 4.0.0, but actual files in 2.5.1. So we need to copy from the source folder instead. But this only affects the libmpdec++ folder. In fact, making this change to the libmpdec folder causes the C header to not get packaged. I don't understand any of why this is the case, but it works like this.
Also to note: I'm just being consistent here with the usage of *_path variables, I know they're deprecated.
There was a problem hiding this comment.
I haven't tested yet, but I'm hoping the weirdness will go away in the followup cleanup PR.
| -MPD_CXXFLAGS = $(WARN) /nologo $(OPT) | ||
| -MPD_CXXFLAGS_SHARED = /DBUILD_LIBMPDECXX $(WARN) /nologo $(OPT_SHARED) $(PGOFLAGS) | ||
| -MPD_BIN_CXXFLAGS_SHARED = $(WARN) /nologo $(OPT_SHARED) $(PGOFLAGS) | ||
| +MPD_CXXFLAGS = $(WARN) /nologo $(OPT) $(CONAN_CXXFLAGS) $(CONAN_CFLAGS) | ||
| +MPD_CXXFLAGS_SHARED = /DBUILD_LIBMPDECXX $(WARN) /nologo $(OPT_SHARED) $(PGOFLAGS) $(CONAN_LDFLAGS) | ||
| +MPD_BIN_CXXFLAGS_SHARED = $(WARN) /nologo $(OPT_SHARED) $(PGOFLAGS) $(CONAN_LDFLAGS) |
There was a problem hiding this comment.
Not sure if this section is still necessary, is this 1.x legacy? If so, I can remove from previous versions too. (there are a few other sections like this too)
There was a problem hiding this comment.
I'm removing these sections in my followup PR
This reverts commit 441adc7.
Ahajha
changed the title
Ahajha
force-pushed
the
ahajha/mpdecimal-4.0.0
branch
2 times, most recently
from
4877d22 to
d7799e1
Compare
Tell me about it! Merges are always tricky - rebasing instead of merging tends to be safer, but solving conflicts can be tricky |
|
(I deleted my comment - bad rebase, in case anyone following along at home is confused :P) What usually happens is that I develop on any of 4 different machines, and when I forget to pull on the one I'm using lots of bad things happen. At least I was able to salvage it this time :) |
|
@perseoGI Can you look at this PR again when you get a chance? Thanks! |
|
@AbrilRBS Can this be merged? |
|
Yup, thanks for your patience @Ahajha! |
|
Thanks @AbrilRBS! |
CPython 3.13 uses this version.
One of the patches related to mingw I believe is no longer necessary:
configure.acnow has explicit checks for mingwAlso,
autoreconfis never called, so these patches are useless anyways. I'll remove them in a future cleanup PR. I've tested locally with Mingw.I also have two other mpdecimal PRs out: #25889 #25903
And one more after all 3 have been merged: #25918