Re: unified versioning RFC (attempt 2)

Alan Knowles wrote:

Thoughts? or do we just allow PEAR to break all the rules and release a|b| etc.

Well I wasnt aware of the problems indicated by Greg quite as clearly. But I did test through all my examples with version_compare() back when I wrote the initial version naming standard. If I understood things properly the main reason why there was opposition to this initial RFC where the following points: 1) manually appending the package state is error prone and inconvinient 2) it is inconvinient to be forced to start at x.0.0alpha1 for packages with a major version number of greater than 1 (as in 2.0.0alpha1 versus 0.1.0alpha) For 1) the obvious solution seems to be adding the package state automatically (as pe rthe rules defined in the original RFC) with some magic determining the proper state number (RC1 -> RC2 etc). For 2) I felt that allowing new major versions to start at 0.1.0 and then having to move to x.0.0 for the first major release is the way to go as in: Foo 0.1.0alpha 1.0.0RC1 1.0.0 Foo2: 0.1.0alpha 2.0.0RC1 2.0.0 Anyways it seems we seriously lack a lawyer because all RFC's seem to have gaping holes here and there (including mine). Its quite difficult to put this kind of stuff into words :-( regards, Lukas