Yes:
>I don't think of a tag that way. Instead, I view it as a name for a
>point in the history of my code.
>
>So one day I may say "I'd like to see the code for "ls" as it was in
>the baselevel tagged "V3.14_baselevel3" and another day I may go
>looking for the code of another piece with that same tag.
>
>To do that, I'd ask Subversion for the subtree in question as it was
>in that rev. Right now I'd use a numeric rev; with aliases I could
>use a named rev.
>
Look, svn-people, all I was suggesting was a practical, simple (I
thought) solution to a
common, simple problem. It's something that all other SCMs offer and it
would appease
many people used to that type of procedure (I think).
As far I can see, all the machinery is already there. The only thing
missing is the symbolic name.
Don't even call it tagging. svn promotes tag-via-copy, fine. Call it
revision naming or aliasing,
whatever. And let the users use it as they see fit. Or not. It would be
an option that wouldn't
bother anyone not using it.
Peace :)
Paul Koning wrote:
>>>>>>"Andrew" == Andrew Reedick <Reedick> writes:
>>>>>>
>>>>>>
>
> >> -----Original Message----- From: Paul Koning
> >> In Subversion, a revision number always identifies a revision of
> >> the repository. The WHOLE repository.
> >>
> >> Any one checkin is likely to change only a subset of the
> >> repository, but nevertheless the rev number identifies a
> >> particular state of the whole repository. If you care what bits
> >> were changed by a particular rev, do "svn log -r 1234 -v" on the
> >> repository root, and svn will tell you.
>
> Andrew> That's my point. The revision number applies to the WHOLE
> Andrew> repository. The Humans use tags to baseline a SUBSET of the
> Andrew> repository. A revision number by itself is useless for that
> Andrew> purpose.
>
> >> So the proposal certainly is logically valid.
>
> Andrew> The OP's proposal is: The svn revision is pretty much
> Andrew> equivalent to what a tag is in other systems. svn alias
> Andrew> <rev> <name> (or svn revalias <rev> <name>)
>
> Andrew> In any system, a tag is a subset of a repository, normally
> Andrew> used to create a baseline. (That's the definition I'm
> Andrew> working under.)
>
>Ok, now I see the problem. Maybe it comes from thinking in CVS, where
>there is no real repository, just a large number of basically
>unrelated files each with its own history and tags.
>
>I don't think of a tag that way. Instead, I view it as a name for a
>point in the history of my code.
>
>So one day I may say "I'd like to see the code for "ls" as it was in
>the baselevel tagged "V3.14_baselevel3" and another day I may go
>looking for the code of another piece with that same tag.
>
>To do that, I'd ask Subversion for the subtree in question as it was
>in that rev. Right now I'd use a numeric rev; with aliases I could
>use a named rev.
>
>Either way, I think of the repository as a 3D file system: the usual
>two dimensions of the name space tree, plus a time dimension. A rev
>and a tag is just a label of a particular value of the time dimension
>-- a plane of the 3D space -- and I can see any part of that plane --
>any subtree in the file system as of that point in time.
>
> paul
>
>
>
>
>
>
>
Received on 2008-02-19 23:58:46 CET