| 1 |
Hi list, |
| 2 |
|
| 3 |
This week I |
| 4 |
|
| 5 |
1. finalized the sci-libs/openblas ebuild file |
| 6 |
and got it merged into ::gentoo. |
| 7 |
|
| 8 |
2. bumped version for sci-libs/blis |
| 9 |
|
| 10 |
3. tried to build all reverse dependencies of |
| 11 |
sci-libs/{blas,cblas,lapack}-reference and |
| 12 |
virtual/{blas,cblas,lapack} to scan for |
| 13 |
possible regression. Result is available |
| 14 |
here [1]. |
| 15 |
|
| 16 |
For some important software I also ran |
| 17 |
unittests, for example numpy and octave. |
| 18 |
|
| 19 |
4. fixed dependency information for two packages |
| 20 |
during the build test. |
| 21 |
|
| 22 |
5. Wrote a preliminary documentation about the |
| 23 |
usage and the project [2] |
| 24 |
|
| 25 |
Next I'm going to integrate intel's math kernel |
| 26 |
library into this mechanism. By the mid-July |
| 27 |
we would be able to present the document to |
| 28 |
more audience and have the mechanism more widely |
| 29 |
tested. |
| 30 |
|
| 31 |
Maybe I'm a little bit ahead of schedule at this |
| 32 |
point as all targets have almost been completed. |
| 33 |
The fact that I'm maintaining a similar mechanism |
| 34 |
for Debian isn't the main reason for this -- It's |
| 35 |
the (unexpected) mechanism redesign that had |
| 36 |
reduced the amount of work. This should be fine |
| 37 |
as the new design is much better and smoother |
| 38 |
compared to the original one. |
| 39 |
|
| 40 |
I talked with Benda about the problem. We think |
| 41 |
Gentoo prefox / portage + Debian integration |
| 42 |
may be the next problem to look into. |
| 43 |
|
| 44 |
[1] https://github.com/cdluminate/my-overlay/blob/master/checklist.txt |
| 45 |
[2] https://github.com/cdluminate/my-overlay/blob/master/usage.md |
Archives