Replying using gcli and Ccing the mailing-list, as the notifications are inconsistent.
AUTHOR : BtbN DATE : 2026-Jul-27 16:22:19 ID : 53293 > Ok? But before pushing this series, everything was working fine. > After pushing > it, fate is broken, on unrelated tests. This needs to be fixed ASAP, > I don't > care about what happened 8 months ago. So far your proposed fix > sounds like No, it was not “working fine”, it was working in an unstudied way and preventing future evolutions, and tests on the top were ensuring that wrong behavior. > something to me that'd take a substantial amount of time to > implement, so it's > not a workable short term solution. It requires: 1. move the code from premultiply filter in a common source file to be called from other filters; 2. use that from vf_scale. I could do it in a few hours, but as I was not offered a part of the money paid to implement this whole mess (and would not want to anyway), I will let Niklas do it. Too bad he did not do it a year ago. AUTHOR : haasn DATE : 2026-Jul-27 16:23:09 ID : 53294 > │ None of this would have happened is Niklas had not pushed twice > over my > │ objections last year or let the matter rot as is. > > None of this would have happened if you actually tested your commits > before > force pushing them to master, intentionally bypassing the CI. I tested my commit, I even have a file with the list of failing tests and checked they all fail because of that exact issue. I pushed like that on purpose. If people prefer FATE to be green while you redo it properly, I am fine with it, I have sent a patch disabling the tests to the mailing lists, but it has not yet happened. > Blaming others for your own mistakes is low even for you. The only mistake on my part was trusting you would do the right thing and wait so long. -- Nicolas George _______________________________________________ ffmpeg-devel mailing list -- [email protected] To unsubscribe send an email to [email protected]
