--- Karl Fogel kfogel@newton.ch.collab.net wrote:
Chia-liang Kao clkao@clkao.org writes:
it seems that cvs2svn hasn't caught up the api change.
Marshall White cscidork@yahoo.com writes:
[appx the same thing]
Applied the patch, thanks guys.
Marshall, note that I didn't apply this portion of your patch:
* subversion/bindings/swig/swigutil_py.c (svn_swig_py_make_file):
Call apr_os_file_put with an argument of O_RDWR instead of O_WRONLY.
We might be wanting to read the file instead of just writing to it.
It's unrelated to the API change, and I'm not sure whether or not
there's a reason for the way the current flags are set up. (Had you
encountered an actual bug related to it?)
-K
Well I *thought* that I did. However I tried it again (without that part of the patch) and it
seems like it worked fine. (I would swear that it didn't work right before...)
This being the case, I agree that there is no reason to apply the other part of the patch at this
time.
I didn't look to see what the APR stuff does with that function parameter. Perhaps nothing.
>From a theoretical standpoint, what svn_swig_py_make_file is telling APR is that the file was
opened write only. The code that later uses the apr_file_t * actually *reads* from the file.
I guess APR doesn't care, but who's to say that it won't later.
Maybe you should at least file the patch away in the we might need later pile.
I'm glad it works now.
Interestingly, it left my repository at a different revision number (236 compared to 238).
Marshall
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 14 02:26:37 2006