Re: [Gnu-arch-users] tla tag

[Tom Lord]

    > From: Zenaan Harkness <address@hidden>
    > > There's some documentation on the Wiki:
    > >     http://wiki.gnuarch.org/moin.cgi/Sealing_20and_20fixing

    > Thanks guys.

    > Here's a question: to me base-0 strongly implies that there could
    > be a base-1, base-2, etc

    > Similarly for version-0, I would certainly expect version-1, etc.

    > Yet this doesn't seem to be the case.

Yes, I *arguably* (not certainly) made a mistake in this area that we
are stuck with until such time as there as sufficient and sufficiently
mature and sufficiently publicly interested attention to arch to fix
it in an orderly way.

The inconsistency is that you know that the namespace-predecessor of
patch-1 is base-0, but you don't know that the namespace-predecessor
of version-0 is unless you look at the archive.

Perhaps it should be version-<K+1> if the namespace-predecessor is
patch-<K>. 

My reasoning at the time was that you should be able to ask for the
"seal" revision by name, without having an archive listing.   You
couldn't do that if you had to guess the K in version-<K+1>.

It's genuinely unclear to me whether I was right or wrong though I do
have my doubts.


    > If the #n doesn't increase, why have it at all? This I've never
    > understood.

I don't call the 0 revision "patch-0" because it isn't necessarily a
changeset revision (one created by "commit" rather than "tag" or
"import").   (Also because, in ASCII, 'b' < 'p'.)

Of course, that's not a very good explanation, is it -- because
`patch-5' isn't necessarily created by `commit' either.

Or maybe it is a good explanation:

   `base' starts a new development line
   `patch' is successive revisions by default, until the development 
     line hits a phase shift to `version'.

Who knows?   As I said: useful structure, but as little definitive
semantics as possible.

Rather, it is up to communities of arch users to form a community of
discourse and agree what these names mean.


    > Finally, that link above uses the phrase "a symbolic version, that
    > is a version where all revisions are created with tag".

I call that a "tag version".

    > So a symbolic version is one where _all_ revisions are created
    > with a tag.

    > I was about to ask if there was a glossary on the wiki,
    > and found it:
    > http://wiki.gnuarch.org/moin.cgi/Glossary

    > That is a really useful thing for me - learning curve and all...

Assume lots of minor mistakes and red herrings throughout the wiki
but, overall, I have the impression it is, in fact, a fantastic
resource for learning arch.  Kind of the ultimate
slightly-wrong-but-basically-good-FAQ++.

-t