Dear Hong Li,
On Tue, 29 Sep 2020 at 16:09, Ludovic Courtès <ludovic.cour...@inria.fr> wrote: > "hong...@mdc-berlin.de" <hong...@mdc-berlin.de> skribis: > > > I had a failure when installing python-umap-learn in guix. It took very > > long time in the 'check' phase. Usually, if "guix weather python-umap-learn" says that there is no substitute, then it is a bad sign. :-) > > The error does not change before and after 'guix pull'. Could you please > > tell me how to solve this problem? Thank you for pointing out the issue. Using the Guix Data Service, https://data.guix.gnu.org/repository/1/branch/master/package/python-umap-learn/output-history one can see the versions which build and the ones which fail (mainly because dependencies, apparently). One the last Succeeded row, you can click on the revision 2020-04-18 19:24:20 which indicates the last known commit where python-umap-learn builds. Therefore, you can run: guix time-machine --commit=cac674d99dc4a332e6210c57ec7f1b8164f66642 \ -- package -i pyhton-umap-learn Hope that helps. > It would seem that one of the tests of python-numba is failing, perhaps > in a non-deterministic way. >From the Guix Data Service, the first offending commit is cac674d99dc4a332e6210c57ec7f1b8164f66642 which updates python-cython. All the best, simon