Re: Re: [PEPr] -1 for RFC::Package CVS/Package Web Access Sync

On Mon Dec 12, 2005 at 02:3307PM +0000, Sergio Carvalho wrote: > Martin Jansen wrote: > >Could it be that you guys misinterpreted Anatoly's proposal? If I got > >him right, he is suggesting that the CVS karma file (on cvs.php.net) is > >automagically synchronized with the karma database table that is used on > >pear.php.net. In practice this means that whenever developers get CVS > >write access to a package, they also become a (lead?) maintainer of the > >package. > > Besides being extremely short for an RFC, I'd say this is a proposal to > give any PEAR developer (anyone with CVS access to PEAR) lead status on > pearweb (the project). Could be -- most likely I'm hitting over my limited English skills. > Your interpretation is something I'd vote +1. It seems bureaucratic to > have pearweb manage permissions and then have developers go ask for CVS > karma to be set manually. Unfortunately, that's not what is on the RFC text. I don't think that this mechanism is a good idea: The CVS ACL system is binary ("write access" vs. "no write access") while the PEAR karma system is more fine grained in the sense that we determine between leads, developers, and helpers. Now if someone gets CVS karma, which PEAR maintainer level should he get? Aside from this issue there is also the problem that not only the database at pear.php.net needs to be updated to reflect the new maintainer, but also the package's package.xml needs to be updated in CVS. (We used to have a system on pear.php.net which made sure that the maintainer database and the package.xml files were in sync, but this has vanished for reasons I never really understood.) And finally implementing something like this requires quite a bit of work because cvs.php.net and pear.php.net are different boxes and thus we'd need to do a fair bit of web servicing etc, which is why the cost-benefit calculation of this whole thing most likely has a negative bottom line. - Martin