On Fri, Feb 19, 2010 at 10:10 AM, Jason Dagit <[email protected]> wrote:
> > > On Fri, Feb 19, 2010 at 9:58 AM, Eric Kow <[email protected]> wrote: > >> On Fri, Feb 19, 2010 at 18:54:41 +0100, Ben Franksen wrote: >> > I can't see any win here over the current system which I personally find >> > perfectly fine and understandable. The 97...99 numbers give clear >> intuition >> > that they are "close to the next version number". The odd/even scheme >> does >> > not have this property. The only advantage of the latter system lies in >> the >> > ability to actually _release_ intermediate development versions for >> > testing. If you plan to do so, well, I'd agree. Otherwise keep it as it >> is. >> >> We do release intermediate development versions for testing (just very >> close to the actual release date). >> >> You also made me realise that this has the drawback of not making the >> distinction between work in HEAD prior to the pre-release, and to the >> pre-release itself. >> >> So I suggest instead a variant of the idea: >> >> darcs 2.5.0.x for unstable work after darcs 2.4 is out >> darcs 2.5.97.x for darcs 2.6 alphas >> darcs 2.5.98.x for darcs 2.6 betas >> darcs 2.5.99.x for darcs 2.6 release candidates >> > > As one of the main complainers about the current scheme, let me throw in my > opinion. > > The thing I find confusing isn't the number or the flip flopping of > odd/even or any of that. It's the duplicity of the names. > > Current we have darcs 2.4 beta 3 which is the same source code as 2.3.99 > (or something close to that). What I find confusing is that one version of > darcs has several aliases. From my perspective, we will continue to have > confusing version numbering until we address this duplicity. > > What I would like to see: One source version, one version name. > > The problem is that when someone says they are using 2.3.98.1 and I'm using > darcs 2.4 beta 1, I don't know if we're using the same version. I default > to assuming we're not. > > So, what I see proposed above doesn't seem to reduce the duplicity. We > still have two names for some versions (2.5.97.x/2.6 alpha). > So, Eric explained to me on IRC that if we went with this last 'hybrid' approach we would try to avoid the alpha/beta/RC terminology and that the only officially accepted version names would be the numeric ones. So 2.5.97.x would be "official" but if someone said 2.6 alpha they are being informal and we shouldn't assume a specific version number to match it. And that he only mentioned the beta/alpha stuff above to bridge terminology. In that light, it does seem that this proposal would reduce to one version name per source version. I'm okay with that then. Or at least, willing to see if it works better than the current system. Jason
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
