2017-01-06: General Discussion¶
Time: 20:00 (UTC)
How many repos? ~1400
How many contributors? ~300
Options for packaging X.org libraries.
* Bundle them together [PR#2068](https://github.com/conda-forge/staged-recipes/pull/2068).
Use build customization to do one build and split out multiple packages.
Drop numpy 1.10 but leave Python 3.4 for a while longer.
John will build Qt 4 on OS X.
Ray will build Qt 5 with jpeg 9* soon on defaults and our problem will go away.
* do not wait for the remaining packages that needs long prefix
investigate why pestc is not uploading
Move MACOSX_DEPLOYMENT_TARGET variable to conda-build-setup and modify the CFEP to reflect that.
pkgw would like to discuss bundling X.org client libraries in the conda-forge stack, as per PR#2068. The PR has everything bundled into a giant tarball for simplicity; gqmelo posted a set of recipes that splits each library out.
Defaults channel libpng and jpeg updates: scheduled for 1Q 2017. Continuum requests closer collaboration on future api/abi incompatible updates to core libraries.
One pinning scheme for both defaults and conda-forge?
That would be ideal. Hopefully we can head that way.
Drop Python 3.4. Now that conda-forge have Qt 4+jpeg9* on Windows the Python 3.5+Windows users can migrate from Python 3.4 to 3.5.
Do MinGW compile things that play well with Python 3.5/MSVC 2015 yet? AFAICT this is still an issue.
Drop numpy 110
Upload of Qt 4 for OS X
Have a VM I’m willing to use to do this. Details here.
Build and upload of Qt 5+jpeg 9* for all platforms
PyCon 2017. Just submitted a place holder talk as we lost the tutorial deadline and the talk deadline is today. Anyone interested in participating please get in touch.
Re-rendering channel improvements. ( conda forge/conda smithy#401 )
conda-build 2 ( conda forge/conda forge build setup feedstock#40 )
Travis CI image change. ( conda forge/conda forge enhancement proposals#6 )
conda-smithy and conda 4.2 ( conda forge/conda smithy#394 )