Use r-base to determine R version(s)

[jakirkham]

Fixes #73

Use r-base (not r) for determining when to fill in CONDA_R.

cc @johanneskoester @mingwandroid

[johanneskoester]

I can confirm that this works for me locally.

[johanneskoester]

Any estimate when this will be merged and become available in conda-forge? We at Bioconda need to proceed with R migration quite urgently.

Thanks!

[johanneskoester]

@jakirkham sorry to bother you, but can we speed this up? Maybe conda-forge could use its own fork of conda-build-all, at least until this has been merged?

[jakirkham]

So I don't have merge permissions here.

Another concern is the tests are failing (though not related to this AFAICT). FWIU PR ( #71 ) should fix this. However, there is a noarch issue that should be sorted one way or another.

[jakirkham]

@pelson and @marqh , could one of you please take a look at this when you are able?

[jakirkham]

Can someone please advise us on how to proceed? Thanks.

[marqh]

Hello @jakirkham @johanneskoester

looking at the history of this repository with respect to continuous integration testing, the story is not a strong one:
https://travis-ci.org/SciTools/conda-build-all/builds
doesn't have any examples of all the tests passing, going back months

with this in mind, I don't see how we can take

Another concern is the tests are failing (though not related to this AFAICT).
as a reason to stop everything, as this appears to be a long standing issue with this repository

given people's desires to make use of this change, I am going to merge this PR, unless I hear strong words to the contrary in short order. The change is clear and comprehensible

@johanneskoester I am not sure of the steps following merging that are required to

become available in conda-forge

Do you know what these steps are please?

thank you
mark

[jakirkham]

Thanks for taking a look @marqh. This plan sounds ok to me.

As far as releasing on conda-forge, we add a PR with the new release to the conda-build-all feedstock and merge once it passes. This will give us new packages for conda-build-all.

That being said, we had run into some funny issues with conda-build-all 1.0.1 at staged-recipes so had to revert to 1.0.0 to fix it. ( #75 ) Would expect PR ( #71 ) should fix this for us, but I'm not sure how you feel about getting that in as well. An alternative might be to revert PR ( #68 ) in the interim. What do you think?

[marqh]

Hi @jakirkham

to summarise, it sounds like you would like

• Use r-base to determine R version(s) #74 merged
• Test get_output_file_path #71 merged
• a new release of conda-build-all

is this correct?

marqh

[jakirkham]

In the ideal case, yes.

[marqh]

ok, one last go to try and get the tests running (#78)

if that fails, we'll go ahead and adopt these changes anyway and get a release out

[marqh]

#78 is still failing, and this is required, so we are going ahead in lieu of test passing

[marqh]

https://github.com/SciTools/conda-build-all/releases/tag/v1.0.2

is now available

@jakirkham @johanneskoester

[jakirkham]

Thanks so much, @marqh.

[jakirkham]

Adding PR ( conda-forge/conda-build-all-feedstock#25 ) to release 1.0.2 in conda-forge.

[johanneskoester]

Thanks a lot!