Re: [llvm-commits] llvm-gcc mirror update delayed again
On Apr 29, 2007, at 7:24 AM, Reid Spencer wrote: > On Sun, 29 Apr 2007 07:11:45 -0700 > Jeff Cohen <[EMAIL PROTECTED]> wrote: >> It happened again last night. According to the log: >> >> - >> --- >> r319 | bwendlin | 2007-04-29 02:04:05 -0700 (Sun, 29 Apr 2007) | 1 >> line >> >> auto-merge 2007-04-29T02:00:00-0700 >> - >> --- >> >> it updated at 2am. But my nightly tester at 3am did not pull in 319, >> only 318 (which it failed to pull in the previous night). But >> updating >> now pulls in 319. > > As I suspected, I think those times are bogus :) > >> >> So when is the mirror updated? If not 2am, why does the log show 2am? >> If it is 2am, then how does an update an hour later fail to see it? > > Bill needs to answer those questions. > I'll need to check, but the process seems to be two-step (I inherited it from Jim L., so I'm not up on all of the fine details). My understanding is that there's a merge script that's run at 2AM every night. That's why you get the "auto-merge" messages with that timestamp. It's from there that another script (which I don't control) does an rsync with the external tree. That one might be coming later in the day. This is yet one more reason why going to the single SVN repository will be *so* welcome. We won't have these types of errors again. >> And >> why does there have to be a mirror in the first place? Why can't >> you do >> your commits directly to a publicly accessible repository? What >> benefit >> is there to keeping the code under wraps for less than 24 hours? > There is no benefit. In fact, it's a needless pain in the ass. :-) > Jeff, we all suffer this frustration. If we could change it, it > would be better by now. It is the way it is because of how Apple > has organized their internal repository. llvm-gcc is comingled with > Apple GCC and that repository isn't accessible outside Apple. > Consequently, the mirror allows llvm-gcc to be accessible by the > community without providing access to the rest of Apple GCC (or > something like that, I'm vague on the details). > > We would have a fully commitable llvm-gcc repository under > Subversion by now if we were cut over, but we're not. Because > there's a lot going on in the community right now, we've put off > the SVN migration until June. So, a few more weeks of patience and > things should get much better in this regard. > I agree. It's a growing pain. Please be patient. An alternative is to take the SVN messages posted here and update your llvm-gcc repository yourself. In truth, the number of changes to the llvm-gcc rep. aren't nearly the number that goes into LLVM proper, so the amount of pain it takes to merge in the patches by hand isn't as great. And it would only be for another month. :-) -bw ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] llvm-gcc mirror update delayed again
Reid Spencer wrote: > On Sun, 29 Apr 2007 07:11:45 -0700 > Jeff Cohen <[EMAIL PROTECTED]> wrote: > >> It happened again last night. According to the log: >> >> >> r319 | bwendlin | 2007-04-29 02:04:05 -0700 (Sun, 29 Apr 2007) | 1 line >> >> auto-merge 2007-04-29T02:00:00-0700 >> >> >> it updated at 2am. But my nightly tester at 3am did not pull in 319, >> only 318 (which it failed to pull in the previous night). But updating >> now pulls in 319. >> > > As I suspected, I think those times are bogus :) > I don't buy it. There are too many days where the changes *are* pulled in one hour later. And other days where they're pulled in 49 hours later. ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] llvm-gcc mirror update delayed again
On Sun, 29 Apr 2007 07:11:45 -0700 Jeff Cohen <[EMAIL PROTECTED]> wrote: >It happened again last night. According to the log: > > >r319 | bwendlin | 2007-04-29 02:04:05 -0700 (Sun, 29 Apr 2007) | 1 line > >auto-merge 2007-04-29T02:00:00-0700 > > >it updated at 2am. But my nightly tester at 3am did not pull in 319, >only 318 (which it failed to pull in the previous night). But updating >now pulls in 319. As I suspected, I think those times are bogus :) > >So when is the mirror updated? If not 2am, why does the log show 2am? >If it is 2am, then how does an update an hour later fail to see it? Bill needs to answer those questions. > And >why does there have to be a mirror in the first place? Why can't you do >your commits directly to a publicly accessible repository? What benefit >is there to keeping the code under wraps for less than 24 hours? Jeff, we all suffer this frustration. If we could change it, it would be better by now. It is the way it is because of how Apple has organized their internal repository. llvm-gcc is comingled with Apple GCC and that repository isn't accessible outside Apple. Consequently, the mirror allows llvm-gcc to be accessible by the community without providing access to the rest of Apple GCC (or something like that, I'm vague on the details). We would have a fully commitable llvm-gcc repository under Subversion by now if we were cut over, but we're not. Because there's a lot going on in the community right now, we've put off the SVN migration until June. So, a few more weeks of patience and things should get much better in this regard. Reid. > >Bill Wendling wrote: >> Have others come across this failure? >> >> -bw >> >> On Apr 28, 2007, at 7:22 AM, Jeff Cohen wrote: >> >>> Once again, the log claims it was updated at 2am PDT, but my nightly >>> tester failed to pull in any changes at 3am. Revision 318 is now >>> present at 7am. >>> >>> ___ >>> llvm-commits mailing list >>> llvm-commits@cs.uiuc.edu >>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits >> >> >> >> > >___ >llvm-commits mailing list >llvm-commits@cs.uiuc.edu >http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] llvm-gcc mirror update delayed again
It happened again last night. According to the log: r319 | bwendlin | 2007-04-29 02:04:05 -0700 (Sun, 29 Apr 2007) | 1 line auto-merge 2007-04-29T02:00:00-0700 it updated at 2am. But my nightly tester at 3am did not pull in 319, only 318 (which it failed to pull in the previous night). But updating now pulls in 319. So when is the mirror updated? If not 2am, why does the log show 2am? If it is 2am, then how does an update an hour later fail to see it? And why does there have to be a mirror in the first place? Why can't you do your commits directly to a publicly accessible repository? What benefit is there to keeping the code under wraps for less than 24 hours? Bill Wendling wrote: > Have others come across this failure? > > -bw > > On Apr 28, 2007, at 7:22 AM, Jeff Cohen wrote: > >> Once again, the log claims it was updated at 2am PDT, but my nightly >> tester failed to pull in any changes at 3am. Revision 318 is now >> present at 7am. >> >> ___ >> llvm-commits mailing list >> llvm-commits@cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits > > > > ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] llvm-gcc mirror update delayed again
Have others come across this failure? -bw On Apr 28, 2007, at 7:22 AM, Jeff Cohen wrote: > Once again, the log claims it was updated at 2am PDT, but my nightly > tester failed to pull in any changes at 3am. Revision 318 is now > present at 7am. > > ___ > llvm-commits mailing list > llvm-commits@cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] llvm-gcc mirror update delayed again
Reid Spencer wrote: > On Sat, 28 Apr 2007 07:22:27 -0700 > Jeff Cohen <[EMAIL PROTECTED]> wrote: > >> Once again, the log claims it was updated at 2am PDT, but my nightly >> tester failed to pull in any changes at 3am. Revision 318 is now >> present at 7am. >> > > Jeff, are you sure you're not dealing with a GMT/Localtime issue? > > In the last few months the synch has always been available around 6-7am PDT, > not 2am. Well, the log shows: r318 | bwendlin | 2007-04-28 02:03:58 -0700 (Sat, 28 Apr 2007) | 1 line auto-merge 2007-04-28T02:00:00-0700 That looks like 2am PDT to me. Also, on most days, it does appear to take place at this time. On some days, it doesn't happen at all. ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
Re: [llvm-commits] llvm-gcc mirror update delayed again
On Sat, 28 Apr 2007 07:22:27 -0700 Jeff Cohen <[EMAIL PROTECTED]> wrote: >Once again, the log claims it was updated at 2am PDT, but my nightly >tester failed to pull in any changes at 3am. Revision 318 is now >present at 7am. Jeff, are you sure you're not dealing with a GMT/Localtime issue? In the last few months the synch has always been available around 6-7am PDT, not 2am. > >___ >llvm-commits mailing list >llvm-commits@cs.uiuc.edu >http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits ___ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits