| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
| |
Also, use the setuptools build_ext command: this allows proper handling of cross-compilation added to setuptools but not to [deprecated] distutils.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Use the return value altogether to check the subprocess execute
successfully or not as in some case it will print some noise
message though run successfully as below.
# python
Python 3.8.10 (default, Nov 26 2021, 20:14:08)
[GCC 9.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import subprocess
>>> cmd = "pkg-config --modversion libxml-2.0"
>>> p = subprocess.Popen(cmd, shell=True,stdout=subprocess.PIPE, stderr=subprocess.PIPE)
>>> stdout_data, errors = p.communicate()
>>> print(stdout_data)
b'2.9.12\n'
>>> print(errors)
b'do_ypcall: clnt_call: RPC: Unable to send; errno = Network is unreachable\n'
|
|
|
|
|
|
| |
2.9.12.
See https://bugs.launchpad.net/lxml/+bug/1928795
|
|
|
|
|
|
| |
libxml2/libxslt. This is needed on Windows but shouldn't get in the way otherwise.
https://www.aleksey.com/xmlsec/api/xmlsec-notes-compiling-windows.html
|
| |
|
|
|
|
|
|
| |
after accidentally renaming them to "--xml2-config" and "--xslt-config" in 4.5.1.
See https://github.com/lxml/lxml/pull/297#issuecomment-640496325
|
| |
|
|
|
| |
Fixes Launchpad bug #1863413
|
| |
|
|\
| |
| | |
Support using pkg-config for xml2-config/xslt-config tools
|
| |
| |
| |
| |
| | |
Those tools are not supported on some distributions where
pkg-config is recommended and otherwise works fine.
|
|\ \ |
|
| | |
| | |
| | |
| | | |
Not linking against libz produces several problems with wheels.
|
|/ / |
|
| |
| |
| |
| |
| |
| | |
If a user is not building libxml2 and is linking to it dynamically, then
there is no additional need to link to libz, since libxml2 will be
depending on libz anyway.
|
| | |
|
|\ \
| | |
| | | |
Include -isysroot linker arg on macos
|
| |/
| |
| |
| |
| |
| |
| |
| | |
Adds -isysroot linker flag when building extensions on darwin target.
Uses `xcrun --show-sdk-path` to determine isysroot value.
Based on suggestions found in https://forums.developer.apple.com/thread/87829
and addresses https://bugs.launchpad.net/lxml/+bug/1727864
|
|/
|
|
| |
provide a build time option so that users can enable it again if they need it.
|
| |
|
| |
|
|
|
|
|
|
| |
modules into the build process and exclude it in PyPy.
Also finally rename lxml.etree.pyx and lxml.objectify.pyx to their correct module names and provide some legacy integration aid by keeping the original API header file names.
|
| |
|
|
|
|
| |
This was needed for Python 2.3, which is no longer supported.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
available
|
| |
|
|
|
|
| |
a build failure
|
| |
|
| |
|
|\
| |
| |
| |
| | |
msabramo/friendlier_error_when_libxml2_not_installed
Friendlier error when libxml2 not installed
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Use distutils.ccompiler.has_function to detect whether libxml2 is
installed in setupinfo.py. This allows failing earlier with a somewhat
more helpful error message if libxml2 is not installed.
On OS X, I display a tip about `xcode-select --install`. Normally I
wouldn't put platform-specific stuff in here (i.e.: I don't advise using
apt-get on Debian/Ubuntu or yum on Fedora/RedHat/CentOS) because it's a
pain to enumerate all the special cases and usually most people on Linux
platforms know their platform well enough to know how to install
packages. I made an exception for OS X and `xcode-select --install`,
because I am pretty savvy with Linux and OS X and getting things to
build, but the `xcode-select --install` thing was not obvious to me at
all. At one point, I had concluded that libxml2 must've been removed
from OS X and I would have to `brew install` it and specify environment
variables to make the compiler find it, because it's a keg-only formula,
but then I found a post on Stack Overflow, which illustrated that
`xcode-select --install` might be all it takes to get Xcode to put the
files in the proper places so that command-line tools can find them.
Since this was not obvious and a good tip, I decided to add it here.
|
|/ |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
With versions below minimum version, instead of raising
an informative error message the following traceback occurs:
TypeError: sequence item 0: expected string, int found
Change-Id: I5ac35935cd61786181e1c6372e752a1600cc2c2d
|
| |
|
|
|
|
|
| |
--HG--
extra : amend_source : 5f766bb41c74b8ea7bba7f71905fb18cb90a19f2
|
| |
|
|
|
|
| |
working code)
|
|
|
|
| |
_Document ref hack
|
|
|
|
| |
strings for names and text instead of byte strings for pure ASCII
|
|
|
|
| |
Extensions through cythonize()
|