"Ian D. Leroux" <idleroux%fastmail.fm@localhost> writes: > On Sun, 15 Mar 2015 11:35:29 -0400 Greg Troxel <gdt%ir.bbn.com@localhost> wrote: >> Certainly it's fair to put a note in the PRs asking the submitters to >> etiher retest with vlc21 or to explain why that isn't a reasonable >> approach. > > Actually, I'm the PR submitter, and I've already tested and reported > that the bug in question has been fixed in more recent vlc versions. > But since these more recent versions are in separate pkgsrc packages, > the fact that they've fixed the bug is irrelevant; the original buggy > multimedia/vlc is still in the tree. OK, but the PR sort of serves as a "I care about this" marker, and it's useful to say that you've moved on. >> My usual request: proposing deletion should explain when the >> to-be-deleted packages were EOLed, or superceded, and why it's fair to >> say to anyone still using them that they are delinquent for not >> updating (that's a bit harsh, but captures the point). [...] > > Your points are well taken, and I should probably not have presented > this as a package-deletion decision. Or at least not initially. The > question I really should have asked is whether anyone on pkgsrc-users > (the listed maintainer) has any interest in multimedia/vlc and is > prepared to help me fix it, or whether everyone has, in fact, already > moved on to more recent versions not suffering from the PR/49447, as I > have. Polling pkgsrc-users is an imperfect proxy for polling the users > of pkgsrc, but I didn't have any better ideas. That's totally fair. Positing "I think there are no users; am I wrong" is entirely reasonable. I'm really reacting to the general notion (not this case) where deletions seem to be overly aggressively proposed. The basic issue is that individual people who keep up have little trouble migrating, and people operating in a semi-production environment have a much harder time upgrading. Probably vlc is not so important in the semi-production environments, compared to Apache/trac/svn/git.... > Put more harshly, should I, or the deeply valued volunteers who spend > their spare time helping me with the PRs I file, invest time and energy > in fixing a package that none of us use anymore? Certainly you should not feel obliged to fix anything just because the package hasn't yet been deleted but instead should do only things that you find worthwhile. > I'm mostly judging from the way firefox has been handled, but at least > for user-facing applications (as opposed to libraries, programming > language interpreters etc.), it makes sense to me that the versionless > package name would correspond to what one would generally install by > default (i.e. latest stable or supported version), and that package > names with baked in versions should be of interest only to those who > have a specific requirement for an unusual older version. I don't want > to have to change the names of the packages I install just to follow > routine progress. > I offer those opinions with all due humility. You're contributing way > more than I ever have, and I do agree with your general position that > package removal decisions should be properly motivated. Consider this > thread preliminary research so that I can decide whether to write a > better package-deletion proposal later. Don't underestimate your contributions; I've been seeing your name for a long time, and I would like to see decisions made based on the merit of the situation, not the reputation of the person suggesting them. I don't mean to come across as "nothing should ever be deleted", and I hope that the "why should X be deleted" message would only take an extra few minutes, when written by someone who is already familiar with the situation. My theory is that this extra few minutes is much less than the time for a few people not familiar to look up the situation. (My current leaning is that you're probably right to suggest deleing the old vlc versions, and also that it's time for python2.6 to be deleted. But the removal of php53, now complete, shows that such things are often more complicated, with a few more packages to be removed as depending on the to-be-removed package, which leads us to ask if those are reasonable to be using, or just cruft.) So please do feel free to post a "let's delete vlc and vlc2" message. It looks like 2.0.1 was in March of 2012, so that more or less argues that 1.1.13 from 2011/12 is pretty crufty, barring reasons why people have to run that in lieu of 2.0.x. But I don't understand the details, and look to you or someone who does to explain it in a few sentences.
Description: PGP signature