Make some ci fixes #279
Make some ci fixes #279
Conversation
git-annex version 8 was previously difficult to install on some systems but now it is default and indeed it is not trivial to install older versions: afni/afni_ci_test_data#3 Forcing a git-annex repository version of 7 is no longer appropriate. It results in a difficult tangle: afni/afni_ci_test_data#2
this reverts previous breakage of coverage functionality. coverage functionality is now turned off for all circleci jobs
these changes should be reverted once the CI test data has been updated... currently the afni_ci_test_data repo has been updated but the files have not been pushed to the afni server
required for thd_incorrelate which makes use of omp... and may not have correctly wrapped code blocks... not sure
This reverts commit 6db3667.
|
Aha, thanks for that @afni-rickr. |
|
It is nice to see the tests pass again, thanks! |
|
Re: "Additionally suma is no longer compiling on MacOS... that error looks more extensive." Does this refer to a CircleCi build error for Mac OS? If that is the case, I think that is resolved---there was two issues that I could tell, each just involving a tweak of "#include .." statements: in one case, for SUMA_input.c, an "#include .." pointed to a package that was just on a local machine (but we that dependency wasn't necessary, so could be fixed just by commenting that line); in another case, for 3dNwarpCalc.c, the recent addition of an if-condition required putting two simple "#include .."s in one branch of the condition. @leej3 , could you please verify if these have resolved the issue(s) you were referring to? Thanks. |
This tries to address the issues in publishing test data with datalad described in afni/afni_ci_test_data#2.
It also includes a breakage in tests that was fixed in #268 but it does not look like that is going to be merged.
A temporary fix for 3dClusterize tests where the output diffing does not occur is added for now to pass the tests until the test data is updated appropriately.
There is still an issue with the compilation of libmri on MacOS described here:
#278
Additionally suma is no longer compiling on MacOS... that error looks more extensive.