Hello everyone. Am Sonntag, 15. November 2009 20:43:50 schrieb Christian Ohm: > On Sunday, 15 November 2009 at 4:23, Zarel wrote: > > "Wait a few days between tagging and actually announcing the release > > publicly" > > Isn't that rule kind of stupid anyway? I thought tags should be touched as > little as possible after tagging, so the waiting period should be before, > not after the tag. And Buginator announced his intention to tag 24 hours > before, so there was some time to test/commit pending stuff/whatever. I do not think that this rule is stupid. But then I wrote it originally, so I may be biased.
It is based on lots of bad experience, and I can also tell you that the "few days" in that sentence certainly has a purpose. What we had once, was "tag, push out src tarball", then the Windows guy would come and "try to build, oh it does not, fix, push out windows build". A little while later someone is building for Mac: "try to build, oh it does not, fix, push out the mac build". Then someone actually tried to run the game on, say, Windows. "Oups, it crashes in level 17 of campaign 40 now, and it is talking about some wrong filename" -> "fix, push out a new build". What we had in the end was 1 tag, 1 tarball, and 3 builds from revisions (or not even that) no one was able to figure out later on. Saying "we are going to tag next week, please build and test" did not help a bit. Announcing the release on the website along with uploading the tarballs made the stuff just worse, because ppl actually downloaded that stuff (who can blame them), and, in case the release contained critical errors and thus had to be retagged, the confusion increased. Thus the rule: "wait a few days between tagging and releasing". I am very sceptic that quality will not suffer if you rely on ppl testing after someone writes an email to the list that he is about to tag 24h later. Not only are not many ppl reading this, they are also too lazy to actually test, especially not quickly, and the actual workers have full time schedules, which do not permit them to throw away whatever they were doing at that moment and just jump into WZ QA instead. > > Download page since this is a beta. But usually, you just remove all > > other downloads from the download page, so for a short period of time, > > no one other than Windows users (and people who compile from source) > > And people who can read and find the SF/GNA links. But no users. Since users usually do not dig into any devstuff sites for links. > On Sunday, 15 November 2009 at 4:44, Zarel wrote: > > Well, there's no reason why Buggy has to handle the entire release, > > Do we have someone else who can make (good) Windows builds? I thought there > were some problems with the crashdumps or something if you don't have > exactly the right setup. There are issues (but they are fixable - don't look at me, I don't have time). If Buginator creates the Windows builds, that would be a nice task already. I think what Zarel means is that he does not also have to carry out the tagging, tarballing, announcements, etc, if that would mean that others cannot get stuff to build and test in time. Regards, devurandom
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Warzone-dev mailing list [email protected] https://mail.gna.org/listinfo/warzone-dev
