[Debian-science-sagemath] 40 (down from 324) failing post-install tests

[Ximin Luo]

Ah, thanks for the heads up. Yes, we should probably also avoid installing that.

But most of the failures was because somehow I ended up with GCC 7 but libstdc++ 6. There were other minor issues to do with our installation tree structure and I've now either worked around them or disabled the tests.

We're back down to 40 failing tests, so I'll call this "good enough" and do a final polish for 7.4-7 which will go into the next Debian stable. Then future work will continue on 7.5.1 etc.

X

François Bissey:
> I’d like to mention that I don’t test the content of sage_setup.
> I don’t install it either, sage_setup should only contain 
> build stuff and no runtime stuff. And I enforce it in sage-on-gentoo.
> If something from there is needed at runtime, I move it somewhere else.
> I have been vocal on trac about that separation and had outburst
> on sage-devel about this.
> In fact I move stuff from “sage” that is install only to sage_setup
> so it is not installed and tested. A lot of the stuff tested in there
> can break on a distro and you definitely don’t want to run it at runtime.
> 
> François
> 
>> On 21/01/2017, at 02:38, Ximin Luo <infinity0 at debian.org> wrote:
>>
>> I will be debugging these this weekend, just a heads up.
>>
>> https://people.debian.org/~infinity0/sage/ptestlong_7.4-6.log
>>
>> sage -t --long /home/infinity0/var/lib/sage/install-test/../sagemath/sage/src/sage_setup/optional_extension.py  # OSError in doctesting framework
>>
>> So they are probably just an issue with how we're installing things into different paths.
>>
>> X
>>

-- 
GPG: ed25519/56034877E1F87C35
GPG: rsa4096/1318EFAC5FBBDBCE
https://github.com/infinity0/pubkeys.git