C. Michael Pilato wrote:
> "C. Michael Pilato" <cmpilato@collab.net> writes:
>>Julian Foad <julianfoad@btopenworld.com> writes:
>>>Look: "svn log" says a file was modified and then added with nothing
>>>in between. That can't be true.
>>>
>>>>~/src/subversion> svn log www/project_tools.html -vq -r13763:14343
>>>>------------------------------------------------------------------------
>>>>r13763 | maxb | 2005-03-30 01:45:59 +0100 (Wed, 30 Mar 2005)
>>>>Changed paths:
>>>> M /trunk/www/project_tools.html
>>>>------------------------------------------------------------------------
>>>>r14341 | maxb | 2005-04-20 22:41:37 +0100 (Wed, 20 Apr 2005)
>>>>Changed paths:
>>>> A /trunk/www/project_tools.html (from /trunk/www/project_tools.html:14339)
>>>>------------------------------------------------------------------------
>>
>>Look more closely. It says it was modified, and then *copied* atop
Yes, I knew that.
>>itself with nothing in between. That's quite possible (though I'd be
I don't think an add (with or without history) can be done over an existing
file without the file being reported as deleted either at the same time or
previously. I may be wrong; that's just a feeling.
>>interested in knowing what hoops the client jumped through to get into
>>that state).
No hoops; the file was in fact deleted in between those revisions.
> Ahem. Let me finish my thought.
>
> Because r14341 was a copy, the history walk on project_tools.html
> jumps back to r14339 (the copy source), bypassing r14340 (in which it
> was deleted). And then it continues back through r13763.
OK, I understand your explanation of why it behaves the way it does. Thanks.
I don't like it though. If I'd asked for the log in reverse chronological
order, its following of history in reverse order would have made more sense.
If the copy had been from one path to a different path, the reason for the
omission would have been more obvious. As it is, I feel I asked for the log of
changes to a certain item and it didn't show me all of them. Oh well, this
isn't as important as a definite bug so I'll leave it at that.
> This is not a bug -- 'svn log' does versioned resource history-based
> reporting, not path-based reporting.
I'll accept that it's not a bug. The reason it behaves like that is not just
because it's doing history-based reporting, but because it's doing it backwards
(yes I know that's the canonical way to do it) even in a forwards log command.
Oh well, that's just how it is. Thanks again for the analysis.
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 8 03:25:59 2005