On 11.06.2009 08:47, Simon Heimberg wrote:
> Am Mittwoch, den 10.06.2009, 21:54 +0200 schrieb Adrian Buehlmann:
>> On 10.06.2009 21:38, Adrian Buehlmann wrote:
>>> What I got was this nice little traceback here:
>> I forgot to mention that I get *no* traceback without Simon's patch.
> 
> I confirm your test results. I agree that a possibly wrong display is
> better than a traceback.

I just tried using CreateFile [1] of windows API instead of fopen

I've pushed my opentest hack to bitbucket:
http://bitbucket.org/abuehl/opentest/src/64cc3eafedbe/opentest.cpp
http://bitbucket.org/abuehl/opentest/changeset/64cc3eafedbe/

If I specify FILE_SHARE_DELETE | FILE_SHARE_READ | FILE_SHARE_WRITE
for dwShareMode of CreateFile,  I can do hgtk thgstatus with your patch
applied while the file is held open by CreateFile.

The temporary file of the atomictemp is then kept as long as
the reading process holds the file handle open (or the process
dies) and then the temporary file is deleted.

It could be that this is the way to go anyway, as I fear we have this
problem anyway too with reading ".hg/dirstate" and there we can't control
the writing side as this is done by Mercurial.

I will look into changing the shellext to use CreateFile instead of
fopen. First for the dirstate reader.

A bit annoying is, we then have to do the file reading using
ReadFile of winAPI ourselves (I can't use the C runtime FILE stuff
or the std C++ lib together with CreateFile)

[1] http://msdn.microsoft.com/en-us/library/aa363858(VS.85).aspx

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables unlimited
royalty-free distribution of the report engine for externally facing 
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
Tortoisehg-develop mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tortoisehg-develop

Reply via email to