On 10/20/06, Jonas Karlsson <[EMAIL PROTECTED]> wrote: > On Fri, 20 Oct 2006 15:17:42 +0200, Hisham Muhammad <[EMAIL PROTECTED]> > wrote: > > > On 10/20/06, Lucas C. Villa Real <[EMAIL PROTECTED]> wrote: > >> On 10/20/06, Jonas Karlsson <[EMAIL PROTECTED]> wrote: > >> > CVSROOT: /sources/goboscripts > >> > Module name: tools > >> > Changes by: Jonas Karlsson <mohjive> 06/10/20 12:34:40 > >> > > >> > Modified files: > >> > Scripts/bin : SandboxInstall > >> > > >> > Log message: > >> > Fix for bu #2, cleaning up target directory when fibo sandbox > >> fails > >> > >> I have been wondering about what to do with the program's contents in > >> this case. The trace file you've attached to the bug shows that there > >> was a binary file installed inside $target/bin. > >> > >> I think that we should take into account the user's choice on > >> "existing entry at /Programs, [r]emove, [k]eep, [w]hatever" when > >> cleaning up the broken installation: if the user has chosen to keep > >> it, just use the fix you've just commited; if the user has chosen to > >> remove it, the fix should also remove $target entirely. > >> > >> Does this look sane? > > > > In any case, that should not be implemented inside SandboxInstall. > > And *please*, let's keep the CVS tree frozen now that we're in release > > candidate status, and apply verified (as in, "patch discussed in the > > list") bugfixes only. As we could see yesterday, even "minor harmless > > commits" can introduce bugs. > > Sorry, but this is my first RC. :)
It's my first RC with CVS, I think :) > I thought that for the final, we should use the same Scipts and Compile as > on the RC unless a severe bug was found, to avoid such problems. Yes. Since there were a number of bugfixes this week, this means there'll be no final next Sunday, but rc2 instead. > I also think that the tree should be branched so that those that don't > work directly on the release can update the tools, without having to worry > about breaking things for the RC. Makes sense. But I was talking to André about switching to SVN as early as next week (one can hope!). This should improve things and make branching easier. > Should all bugfixes go through the mailing list during rc status? What > about this fix? We never had well-defined rules on this. But I think it's a good one to have. Does everyone agree? -- Hisham _______________________________________________ gobolinux-devel mailing list gobolinux-devel@lists.gobolinux.org http://lists.gobolinux.org/mailman/listinfo/gobolinux-devel
