Nice quoting of p5git://patchlevel.h. But I stand by my assessment of that usage as unusual, even strange.

"Perl version 6" finds ~55k references for something I expected to be a quite common usage. Since, according to you, Your Mother, and patchlevel.h there actually aren't any versions of "Perl revision 6" (well, it seems premature to me, anyway), I'm pretty sure that google isn't actually finding (many) instances of "Perl version 6.0" (4 isn't "many", especially since 3 of them are actually talking about "Perl subversion 6.0.0" or something even more specific).

By (extreme) contrast, "Perl revision 6" finds 6 hits (7 by now, surely) and it looks like at least half of them aren't even talking about "revisions of Perl" as you seem to think we should. :)

I also expect it to be quite common to read something like "Perl version 5.6.1"[2M] (and not to say "Perl subversion 5.6.1"[0!]).

So, the internet appears to know that "Perl version" applies to things like "6" and like "5.6.1", the two positions where the term shouldn't be applied. I'm not sure how to use google to prove that when people write "Perl verison 5.6" they usually are just using it as shorthand for "Perl subverison 5.6.0" and not for "Perl subversions from 5.6.0 but before 5.7.0", but I really believe that they usually are.

After meditating on the definitions of "revision" and "version", I can support the use of "revision" for things like "Perl 6" from a linguistic point of view. But the internet proves that such usage wouldn't be adopted (patchlevel.h has used those terms for a very long time). So I'd probably go with something more like version.genus.release (lots of synonyms for "genus" could be substituted but "genus" is my favorite so far) but realize that "major version" or "major revision" is probably required for clarity (and realize that "genus" still wouldn't be clear just because nobody much talks about such things).