Conversation
|
I don't get the failure of the CI (the one with opam). It looks like it is trying to install analysis 0.2.3... |
|
@affeldt-aist I've just seen your comment; maybe this is due to the fact the line Line 25 in fac1196 does not mention a version, so opam takes the first version it is aware from coq-released (0.2.3) so maybe you could try replacing it with this: ? |
|
or alternatively, add |
I thought |
|
yes but if no version is mentioned, the algorithm is a bit non-intuitive (it does not defaults to dev…) (There had been a related discussion in a coq issue, I'll try to re-find it just FTR) |
|
OK, the discussion I was thinking about is this PR: which quotes the following comment in the
(but I don't know if this is also documented in the OPAM manual) |
|
(I pushed the fix suggested by @erikmd but now the CI does not trigger the check using opam anymore...) |
Actually according to this page it seems the build was indeed triggered (but the status feedback was lost on github side): https://travis-ci.org/github/math-comp/analysis/jobs/696872899 and the error is now different: |
|
I see. Since |
Well, we should not start making packages from the leafs of the dependency graph 😆. |
|
The new error is still puzzling: There is no |
Yeah, the general problem seems to be the wrong codebase is pulled. |
|
It looks like it is pulling 4b3281e. |
a88d7ff to
e562b94
Compare
Patch for math-comp/math-comp#516