On Tue, May 27, 2008 at 10:57:18AM +0200, Branko ??ibej wrote:
> Bernard Blackham wrote:
>> I had to use the APR functions for detecting errors, as the previous
>> errors were Win32-specific. Using APR_STATUS_IS_EACCES actually
>> tests for many more errors than previously. This might hurt
>> performance a little - I tried to measure it, but my checkout times
>> vary quite wildly and any differences are lost in the noise here.
>>
>
> I'd really prefer to have the error checking part special-cased fro
> Cygwin and Windows, so that the set of codes tested on Windows doesn't
> change.
Thanks for the feedback Branko.
I've respun the patch to preserve the Win32 error codes and added a
bunch of error codes that mirror the Win32 ones (so I believe).
FWIW, EPERM is the only one I've seen ("Permission denied" is
printed). I don't know what other versions of Cygwin (or underlying
Windows versions) return, so I've taken a cautious approach and not
pruned the list at all.
> I'm not so worried about the performance regression since it
> would only manifest itself in cases where the operation would fail now;
Not quite true - as I read it, the error codes are checked against
this list after every operation, successful or not (to determine
that it was infact successful).
Anyway, new patches are attached for trunk/1.4.6 and 1.4.5 again. As
before, I've only tested the 1.4.5 patch (it's still in the middle
of a multi-gigabyte checkout, but it would usually have failed by
now :)
Kind regards,
Bernard.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-05-27 12:23:57 CEST