| <<<Back 1 day (to 2016/04/02) | 20160403 |
RobinWattsLenovo | tor8: When we use multiple threads (multiple contexts) in MuPDF, we leak 1 style context per context. | 04:55.43 |
| Should style contexts be shared? If so, they probably need to be protected with a lock. | 04:56.11 |
sebras | marcosw: hm.. how does this regression testing decide what commits to test? | 09:12.11 |
| marcosw: the commit for MuPDF tested in the latest e-mail, d8b0666, seems to be quite old..? | 09:12.44 |
| marcosw: I guess it is just me that doesn't yet fully understand how this works though. :-? | 09:13.21 |
marcosw | sebras: the nightly regression testing for mupdf tests up to 5 commits per night. I do this because the mupdf commits tend to happen infrequently but in large groups. | 14:26.21 |
| sebras: If I tested all of the commits at once and there were regressions it would require more work to isolate the responsible commit. | 14:26.41 |
sebras | marcosw: I see. | 14:29.49 |
| Forward 1 day (to 2016/04/04)>>> | |