Philip Martin wrote:

> Julian Foad writes:
>>  Philip Martin <philip.mar...@wandisco.com>
>>>  vijay <vi...@collab.net> writes:
>>>>   Attached the patch and log message.
>>> 
>>>  Committed in r1418322. Thanks!
>> 
>>  Hi Philip and Vijay.
>> 
>>  Your log message:
>>  [[[
>>  * subversion/libsvn_ra_serf/replay.c
>>    (svn_ra_serf__replay, svn_ra_serf__replay_range): Replace the
>>      session url string with the request path portion of the url.
>>  ]]]
>> 
>>  leaves me thinking, "But *why* did you replace the session url string 
> with the request path portion of the url?"
>> 
>>  Please edit the log message to say why this change was made and what
>> it means in functional terms (like, what behaviour does it change,
>> does it fix a bug, etc.).
> 
> I don't think a log message is the place for a detailed explanation of
> why a member called 'path' should contain a path rather than a full URL.

I absolutely agree.

> That's the sort of documentation that should go in the code.  I suppose
> a log message like: "Set the handler's path member to a path instead of
> a complete URL" might be better.

That would be a little bit better because it gives a hint that it is correcting 
some sort of error.

But isn't the subject line of this email a relevant clue to what's happening?  
I don't 100% grok it, but it sounds like the issue being adressed by this patch 
must be something like:

  The mod_dav_svn replay report was logging the entire session
  URL in the httpd access log.  Make it do what it is supposed
  to do, which is log just the path-within-repos.

No?

(Please sanity-check that because I'm just guessing.)

- Julian

Reply via email to