On Thu, 2019-01-03 at 21:39 +, James Le Cuirot wrote:
> The previous approach would erroneously match foopython. The new
> approach requires the match to start the string or be preceeded by a
> slash, the only two cases we actually want. It does this with slightly
> less code and allows the rep
Hi James,
James Le Cuirot writes:
> Once this is in place, I can finish my long-awaited revamp of my
> cross-boss project that will allow you to cross-compile @system from
> scratch with very little effort.
I haven't gone through the patches yet. But I want to say thank you!
The cross-boss pro
Following the changes in python-utils-r1.eclass, Python is now needed
in both BDEPEND and DEPEND.
As distutils does not call our eclass helpers, we need to pass the
correct include and library directories in setup.cfg. Unfortunately it
is still hardcoded to add -I/usr/include/pythonX.X and -L/usr/
It is useful on its own when the build system calls setup.py for us,
from a Makefile, for example. ${BUILD_DIR} is unlikely to be set in
this instance so it falls back to ${PWD}.
Signed-off-by: James Le Cuirot
---
eclass/distutils-r1.eclass | 15 ---
1 file changed, 8 insertions(+),
This previously happened before checking but now exporting PYTHON
involves creating the wrappers, which requires the requested Python to
be installed.
We could also drop all the calls to python_wrapper_setup as this is
now already done when exporting PYTHON but it doesn't hurt to be
explicit.
Sig
python-any-r1 is not to be used in packages requiring Python at
runtime. It may not even be installed.
Signed-off-by: James Le Cuirot
---
eclass/python-any-r1.eclass | 5 +
1 file changed, 5 insertions(+)
diff --git a/eclass/python-any-r1.eclass b/eclass/python-any-r1.eclass
index 7c5ae8787
We assumed Python to be located at /usr/$(get_libdir) in recent
changes to this eclass but this has changed from Python 3.7. We now
follow upstream, installing most files to /usr/lib, with only certain
files, such as the pkg-config files, going into /usr/$(get_libdir).
Signed-off-by: James Le Cuir
dev-lang/python sometimes used to install the epython module into the
wrong libdir (e.g. lib vs lib64). The earlier eclass changes actually
break the build entirely as sysconfigdata is now sourced from within
SYSROOT, where it may not even be installed yet. This therefore needs
to be handled as a s
Shebangs may need fixing on prefix systems or when cross-building but
not at other times.
Signed-off-by: James Le Cuirot
---
eclass/python-utils-r1.eclass | 8 ++--
1 file changed, 2 insertions(+), 6 deletions(-)
diff --git a/eclass/python-utils-r1.eclass b/eclass/python-utils-r1.eclass
ind
This will deal with cases where the shebang is defined using ${PYTHON}
or the location of python in the PATH.
Signed-off-by: James Le Cuirot
---
eclass/python-utils-r1.eclass | 22 --
1 file changed, 20 insertions(+), 2 deletions(-)
diff --git a/eclass/python-utils-r1.eclass
Python has little concept of cross-compiling but it turns out that
making it work isn't so hard after all.
Platform-specific details are held within _sysconfigdata.py,
sysconfig.py, and various distutils files. If we simply symlink these
from SYSROOT into an empty directory and add that directory
The previous approach would erroneously match foopython. The new
approach requires the match to start the string or be preceeded by a
slash, the only two cases we actually want. It does this with slightly
less code and allows the replacement of whole path strings that would
be problematic when pass
It currently defaults to EPYTHON, which might not even be defined
yet. The "impl" arg defaults to EPYTHON anyway.
Signed-off-by: James Le Cuirot
---
eclass/python-utils-r1.eclass | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/eclass/python-utils-r1.eclass b/eclass/python-uti
If Python returns relative site or script directories then something
is very wrong!
Signed-off-by: James Le Cuirot
---
eclass/python-utils-r1.eclass | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/eclass/python-utils-r1.eclass b/eclass/python-utils-r1.eclass
index e3c
Here's a series of eclass changes I've been working on since August to
allow Python modules to be cross-compiled. We previously believed this
to be practically impossible without significant changes upstream and
in the wider Python ecosystem so getting here feels like quite an
achievement. While th
On Thu, 3 Jan 2019 00:25:29 +0100
Jonas Stein wrote:
> git log | grep "Author: "| sort | uniq | sed "s/Author: //g" | wc -c
That's a rather round about way of doing :
git shortlog -e -s | cut -f 2
git shortlog -e -s | cut -f 2 | wc -c
37471
git shortlog -e -s | wc -l
998
pgpFfkBYRuopC.
On Wed, 2 Jan 2019 13:58:26 -0800
Georgy Yakovlev wrote:
> +# @CODE
> +# ECARGO_BUILD_FLAGS="$(usex pcre "--features pcre2" "")"
> +# @CODE
You may also want to provide user's hooks as well.
Similar to EXTRA_ECONF and EXTRA_EMAKE those might be
ECARGO_EXTRA_BUILD_FLAGS
ECARGO_EXTRA_INS
Dnia January 3, 2019 5:46:35 AM UTC, Georgy Yakovlev
napisał(a):
>On Wednesday, January 2, 2019 8:52:45 PM PST Michał Górny wrote:
>> On Wed, 2019-01-02 at 13:58 -0800, Georgy Yakovlev wrote:
>> > This adds 2 eclass variables
>> >
>> > ECARGO_BUILD_FLAGS
>> > ECARGO_INSTALL_FLAGS
>> >
>> > cont
18 matches
Mail list logo
