On Monday, 2013-03-25, dE . wrote: > As stated before, the best way to find bugs is constant usability testing. > Beta 1 and beta 2 have a few days in between releases, so what do you > expect these testers to upgraded every day and yet use their system > normally? Instead they attempt to just see if everything works externally, > and upgrade to the next beta. The same happens with RC releases. These > frequent releases cause confusion. By the time you find a bug for RC1, RC2 > is already out, and the devs in the bugzilla will ask for you to upgrade. > Who has that much of patience? The time between those test milestones is about two weeks. This is a compromise between convenience (packaged software) and relevance (not encountering issues already fixed in the mean time). Of course, ideally there would be tons of people running master, but experience shows that way more people are likely to test drive packaged software. So at some stage during the development cycle, a series of packaged milestones is offered for those who prefer this form of software distribution for testing. At some point it might be possible to do daily packages, but we haven't arrived at that yet. > If you want constant usability testing, the target userbase should be > somewhere between devs and end users; these people want a usable system, > and don't mind testing. But for that to happen the releases should be slow, > so it reaches them and it's convenient to them. But that's already the case, isn't it? It is basically up to each individual to chose their point of involvement. From running git pull --rebase every couple of minutes to upgrading packages every couple of months. The latter depending on the package source, varying between weeks (e.g. rolling release distro) to several years (LTS distro). > Also real testing starts with RC after the freeze -- which ensures no new > bugs. But unfortunately, there's not even a month between RC3 and the > freeze -- how do you expect to find new bugs in a few weeks? Are you sure? For 4.10 I see the freeze happend on November 8th, R3 tagging on January 17th. About two months if my math doesn't fail me. > Instead there should be 3 or 4 months between RCs, so bug can be collected > and the RC releases reach across layers of users; it shouldn't happen that > before the release reaches the user, it becomes outdated by 2 other > releases. But wouldn't longer periods between RCs not mean that people encounter the same bugs over and over again instead of finding new bugs due to already known ones having been fixed? Cheers, Kevin -- Kevin Krammer, KDE developer, xdg-utils developer KDE user support, developer mentoring
Attachment:
signature.asc
Description: This is a digitally signed message part.
___________________________________________________ This message is from the kde mailing list. Account management: https://mail.kde.org/mailman/listinfo/kde. Archives: http://lists.kde.org/. More info: http://www.kde.org/faq.html.
