Issue with latest build about boost-cpp-feedstock HOT 8 CLOSED

Probably needs a zlib in host.

from boost-cpp-feedstock.

I'm seeing the same issue with my feedstocks. Agreed, looks like zlib was pulled in implicitly before but not any more (doesn't look like PR #113 was at fault to me, maybe something else changed in the meantime).

The latest build reports zlib: no in Boost's configure output (at least on linux_64); the previous build reports zlib: yes in Boost's configure output.

from boost-cpp-feedstock.

More specifically, the boost::iostreams module was accidentally compiled without the zlib filter. I'm happy to contribute a PR to fix this, but I'd like to ensure that it stays fixed, since it crops up from time to time (e.g. in defaults ContinuumIO/anaconda-issues#414, ContinuumIO/anaconda-issues#9265). The most robust way to do that IMHO would be to compile and link a little C++ test program as part of the test suite, but I'm sure there are subtleties to making that work cross-platform. I can trawl through some feedstocks to see if I can find others where this is done, but if you know of any off hand, that would be most helpful!

from boost-cpp-feedstock.

I wonder if it would be possible to force boost to try to build with a specific configuration. Then the build would fail if it was missing a library it needs for that configuration. That would avoid silent config changes like this.

from boost-cpp-feedstock.

This should be resolved now that #115 has been merged.

from boost-cpp-feedstock.

It looks like in the latest builds, zlib has become disabled again.
For example https://dev.azure.com/conda-forge/84710dde-1620-425b-80d0-4cf5baca359d/_apis/build/builds/489010/logs/27
this shows zlib: no.
The linux-64/boost-cpp-1.77.0-he72f1d9_2.tar.bz2 build published 10 days ago is broken, whereas the linux-64/boost-cpp-1.77.0-h359cf19_1.tar.bz2 build published many months earlier is OK.

from boost-cpp-feedstock.

I might be reading this wrongly, but it seems like #115 was not actually merged into main? for example, the recipe/test folder which tests for this failure isn't present on main.

from boost-cpp-feedstock.

@kcajf please file a new issue with all the information requested in the template. Thanks!

from boost-cpp-feedstock.