From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chris Faylor To: Jason Molenda Cc: overseers@sources.redhat.com Subject: Re: Controlling cvs commit access Date: Mon, 16 Oct 2000 16:45:00 -0000 Message-ID: <20001016194459.B3344@cygnus.com> References: <20001014002746.A5942@shell17.ba.best.com> X-SW-Source: 2000-q4/msg00026.html Message-ID: <20001016164500.P1DZ_PmDmDIbxaE_btWhvXAwFtR0R4beu_v2CmQ3Mj0@z> On Sat, Oct 14, 2000 at 12:27:46AM -0700, Jason Molenda wrote: >We have a large engineering organization at Yahoo with a wider >variety of cvs uses than I'd seen before. I needed some fine-grained >way of controlling access to parts of the repository, so I wrote >a nice little access control list script. It lets you express >things like > > User FOO may only check in changes in directory BAR > > Only users FOO, BAR, and BAZ may check in changes in directory BOO > > User MOO may not check in any changes to BOO > >And all the other variants I could think of as being useful. There >is a simpler access control script included in the cvs contrib dir >(cvs_acls.pl), but it sacrifices the expressiveness that my script >has for much greater simplicity of implementation. > >I don't know if this thing would be useful to anyone outside Yahoo, >but if, for instance, there was ever a hope of merging >gcc/gdb/binutils/newlib/etc and people were concerned about >controlling commit policy, it would be easy to do with this. > >Lemme know if y'all want to use it; I'll do a little cleanup and >send it along. It's being used at yahoo by a large developer base, >I'm confident that it's been well debugged by now. I haven't seen anyone reply to this, but I would say that this is very interesting and I'd welcome a patch or even a patched CVS. cgf