On Sat, 15 Nov 2003, Stas Bekman wrote: > Randy Kobes wrote: > > >>Yeah, well, the problem is that now there's no 1.99_11 binary on > >>Winnipeg :( :) > > > > At one point I tried keeping multiple versions. This becomes > > a bit complicated administratively, but more importantly, > > 99.9% of the time older versions have bugs that are fixed in > > the current version. This is the main reason I use the cvs > > version for the builds - however, I make sure that all, or > > at least the vast majority, of tests pass with a given > > build. So there's rarely a case of needing an older version, > > especially for something like mod_perl 2 which is in > > development. > > May be we should keep the binary of the latest release and > the cvs one? The cvs version may introduce bugs > non-existing in the last release and may go unnoticed for > quite some time, if 'make test' doesn't catch them. So if > winFU is now a viable platform for production use of > mod_perl 2.0 users will most likely be better off with the > latest release and not the cvs version. > > I'm not sure how ppm can handle that, though. Can it > ignore the dev versions by some method similar to PAUSE (_ > in the version number)?
That's a good point about the cvs vs last release ... ppm just downloads the name specified by the package, so we could make up two packages - "mod_perl" (the current official release), and "mod_perl-cvs", say. -- best regards, randy --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
