Blas binding should be even better at recognizing compatible memory layouts #340

Activity

1. lloydmeta commented on Aug 15, 2017

   

   lloydmeta

   on Aug 15, 2017

   Contributor

   More actions

   Just copying what the concrete todo we found in #190 is:

   needs to ignore the stride of the axis that is only length 1

   I'm kind of new to the whole linalg lib thing but it seems like I can try to tackle this one (while the code is still fresh in my mind), if that's ok with you?
2. bluss commented on Aug 16, 2017

   

   bluss

   on Aug 16, 2017

   MemberAuthor

   More actions

   Sure, go ahead. I wouldn't mind some fresh eyes on the blas binding stuff, it's pretty ad-hoc. Feel free to discuss with me on irc in #rust-sci or somewhere else; I might be there if we are lucky enough to be online at the same time.

   ndarray has stubbornly turned of it's library-embedded tests ("unit tests") in favour of the tests/ directory but we might want to have embedded tests for this. Depends on the best way one finds for testing it.
3. lloydmeta

   added a commit that references this issue on Aug 17, 2017

   Adds blas support for transposed row matrices (Closes rust-ndarray#340)

   ...

   This issue will close once commit c1958d2 is merged into the 'master' branch.

   c1958d2
4. bluss

   closed this as completedin c1958d2on Aug 17, 2017
5. bluss

   added a commit that references this issue on Aug 17, 2017

   Merge pull request #342 from lloydmeta/enhancement/blas_row_major_2d

   ...

   This issue will close once commit 0b4cdd4 is merged into the 'master' branch.

   0b4cdd4