Using new compilers on linux fails with "Failed to import the site module" #2523
Comments
|
Could you try: |
|
Thanks! That seems to fix it. Does it mean that the |
|
The critical thing is that you must have defaults Python in your root
environment. It might be best to keep conda forge out of your root
environment, but add it into other envs. Each environment can have its own
.condarc file.
The need for this will go away over time, I think, as conda forge adopts
conda build 3 and the new compilers.
…On Nov 25, 2017 13:51, "Zaharid" ***@***.***> wrote:
Thanks! That seems to fix it. Does it mean that the defaults channel has
to have higher priority than conda-forge for the compilers to work? Is it
possible to use both channels?
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#2523 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AACV-c_HDb0njHrNj6WGbk7osB6YVuIYks5s6G_TgaJpZM4QqiN8>
.
|
|
Sorry that's not quite accurate. Yes, right now, if conda forge is higher
priority than defaults, building Python extensions with conda build will
not work. There is extra stuff we include with Python to make it work, and
if conda build create build envs with conda forge Python builds, that stuff
will not be present.
…On Nov 25, 2017 14:07, "Michael Sarahan" ***@***.***> wrote:
The critical thing is that you must have defaults Python in your root
environment. It might be best to keep conda forge out of your root
environment, but add it into other envs. Each environment can have its own
.condarc file.
The need for this will go away over time, I think, as conda forge adopts
conda build 3 and the new compilers.
On Nov 25, 2017 13:51, "Zaharid" ***@***.***> wrote:
> Thanks! That seems to fix it. Does it mean that the defaults channel has
> to have higher priority than conda-forge for the compilers to work? Is
> it possible to use both channels?
>
> —
> You are receiving this because you are subscribed to this thread.
> Reply to this email directly, view it on GitHub
> <#2523 (comment)>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/AACV-c_HDb0njHrNj6WGbk7osB6YVuIYks5s6G_TgaJpZM4QqiN8>
> .
>
|
|
Somehow your suggested workaround does not work anymore @msarahan. For reference see https://travis-ci.org/markovmodel/PyEMMA/jobs/438821050 the root env is kept in the state from miniconda3 installer, and then the compiler is being dragged in by conda-build into the build environment. Is there any chance to make this work (again, as it used to work eg. last week or so). Thanks a lot for looking into this. |
|
I believe this is a conda-forge issue, as discussed at conda-forge/python-feedstock#203 I'm closing this issue, because I believe it is unrelated to conda-build. Please open a new issue if you would like to explain why you disagree. |
|
sorry for using this issue kind of off-topic. |
|
Hi there, thank you for your contribution! This issue has been automatically locked because it has not had recent activity after being closed. Please open a new issue if needed. Thanks! |
I want to update this recipe:
https://github.com/Zaharid/conda-recipes/tree/master/lhapdf
to use the new compiler infrastructure. I rewrite the
meta.yamlfile this way:I get the following error when trying to build the recipe on a linux (Mint 18.3) laptop:
I am not sure where the error comes from. Seems to be some activate script. Note that it seems to be looking at the
pythonexecutable in the root environment rather than the one that gets installed.Full output:
The text was updated successfully, but these errors were encountered: