New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Boost unification #164
Boost unification #164
Conversation
Since py35 and py36 both use the same vc version, it's not necessary to build boost-cpp in both versions. This disables py36 builds.
Disable python 3.6
MNT: Re-rendered with conda-smithy 2.0.0 [ci skip]
Remove Python headers
Fix checking for non-existence of python headers
Update ICU pin
…render_master MNT: Re-render the feedstock [ci skip]
Update to 1.64
* Boost ticket: https://svn.boost.org/trac/boost/ticket/12516 * Patch taken from upstream commit 1d86261: boostorg/serialization@1d86261
Fix missing include in boost::serialization
Update to 1.65
Update to 1.65.1
I added data about infixes too. Most of the devel stuff is either a CDT or a Perl package, but I guess that counts too. That said, you are right, and this is conda-forge-wide issue, not just Boost. Maybe worth discussing naming patterns separately: what our conventions are and perhaps even more importantly how to enforce them. |
Co-authored-by: Marcel Bargull <mbargull@users.noreply.github.com>
…nda-forge-pinning 2023.09.20.07.15.04
We discussed the naming in the core call today, seems that everyone's fine to move forward with |
Alright, it's been almost a week, let's get this in. Thanks everyone! |
Implementation for conda-forge/boost-cpp-feedstock#137
Closes #53
Closes conda-forge/boost-cpp-feedstock#25
As discussed in that issue, this PR:
boost-cpp
->libboost
&boost
->liboost-python
(maintaining the old output names as wrappers for now)libboost
(whileboost-cpp
stays unchanged without one)libboost-headers
that will fill the role of the run-export-lessboost-cpp
The history of https://github.com/conda-forge/boost-cpp-feedstock is preserved through a merge (after some preparatory commits to reduce collisions). This makes the PR look unwieldy but the changes before the merge aren't very interesting. The upside is that this allows digging into the full archeology in one place (with the only snag that GH now resolves the PR links incorrectly).
Implementation notes:
libboost
depends onlibboost-headers
, so the headers are really only packaged onceb2
built tool doesn't seem to support only-build-don't-install so well, and more importantly, because we need to buildlibboost-python
per python version.libboost
&libboost-headers
are built only once inbuild.sh
/bld.bat
(essentially the build scripts from boost-cpp), with the only substantial change that they're now "installed" into a temporary prefix from which we can then actually install the respective files into our $PREFIX per outputlibboost-python
essentially keeps the build scripts of boost, only that we now reuse theb2
that was bootstrapped during the global build phase (this now allows full cross-compilation; previously ppc was emulated here).build-py.{sh,bat}
and the global build scripts, but for now I've tried to keep the changes to the respective histories to a minimum.libboost-python
to only package dynamic libs, and add a run-export. Also, while it previously depended on the fullboost-cpp
, the link-check says the libs aren't necessary, so switch to depend onlibboost-headers
instead.PTAL @conda-forge/boost @conda-forge/boost-cpp
CC @conda-forge/core