> On Wed, 2014-01-08 at 05:09 -0500, Adam Jensen wrote: > > On Wed, 8 Jan 2014 09:26:50 +0100 (CET) > > [email protected] wrote: > > > I think we don't really have the resources to maintain 2 or 3 > > > branches. > > I suspect we need two : release and current. More than that ... I > would agree...
Right. That's what we do currently. > > Hmm, can either of these methodologies tolerate non-trivial > > development? > > For example, what if a developer(s) wished to extend GHDL such that > > complex data structures could be traced? > > The usual Mercurial approach to "big picture" changes is to clone the > entire repo (see www.hginit.com, last chapter) essentially forking > the > project for the separate development. It's safer (there is no danger > op > polluting the main repo) and the changes can later be merged back > into > the trunk (or a separate new repo if preferred) Yes. Experimental work shouldn't be issue with mercurial. > I'm not yet clear how that plays with Sourceforge but the "fork" > button > appears to work along these lines (making you a repo under your > account) > where you can play, and it does allow merge requests back to the > original (which an admin here would have to perform!) > > I'm trying this out with another SF project but haven't had time to > learn it properly yet. > > > > I also have my private test suite, and I will test GHDL on it. > > private? > > It may contain testcases which Tristan cannot make public domain for > whatever reason... There are: * designs from various sources, that doesn't constitute a testsuite (i.e. they weren't written for that) and I think they have no place in the repo * old bug reports (but the redistributable status isn't clear) * tests I have found on the net (again, status isn't clear) Regards, Tristan. _______________________________________________ Ghdl-discuss mailing list [email protected] https://mail.gna.org/listinfo/ghdl-discuss
