"Greg Stein" <gstein@collab.net> writes:
>
> Why should people do a fresh checkout? Wouldn't the only problem be that the
> keyword substitutions are a bit "off"? I certainly don't mind.
Isn't it kind of bad to have blatantly wrong keyword info attached to
a file's entry?
> Or will the substitution *require* that created-rev be in the
> entries file; therefore, we must do a checkout to ensure that it
> gets there?
I'm sure that we'll do graceful degradation. If a keyword value isn't
in the entries file, then it will substitute to an empty string.
But beyond that: we'd eventually like 'svn status' to make use of the
info, even if very few of us ever use keyword expansion. Someday I'd
like to have 'svn st' tell me the CR of a file, so I can immediately
update the file to revision CR-1. (We had to do that today, in fact.)
Someday I'd like status to tell me the last person who changed a file
too. Sure, it's all doable via 'svn log file', but it's nice to have
alternatives.
The point about keyword substitution: if you checkout a working copy
from a client/server combo later than r739, you are guaranteed that
the keyword info in your working copy will always be correct, and that
it will always be maintained correctly. If your working copy is
missing info, or if you update from a pre-r739 server, then you lose
that guarantee. As long as svn developers are willing to live without
this guarantee, that's fine. But they should be aware of the
tradeoff. (I mean, a fresh svn checkout isn't that big a deal.)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:55 2006