[python-win32] Overlay Icon Question and network drives

[Jürgen Kareta]

Tim Golden schrieb:
> python at kareta.de wrote:
>   
>> Hi Andrea,
>>
>> the code in my first link was from the mercurial extension TortoiseHG. They
>> found some issues that the icon overlay slows down network devices. See here:
>>
>> https://sourceforge.net/tracker/index.php?func=detail&aid=1892791&group_id=199155&atid=968354
>>
>> Could you please tell if you have the same issues with the code Roger posted ?
>>     
>
> Worth noting that *any* code which is messing with icon overlays should
> be just about as tight as it can possibly be. TortoiseSVN, which is written in 
> C++ has been doing this for years, tweaks its overlay & cache code on 
> pretty much every minor release to squeeze some more performance.
>
> I had a quick glance at the TortoiseHg code and was a bit surprised that
> the key _get_state function isn't as tight as it might be. I say that, however,
> with a huge amount of diffidence as I've had nothing whatsoever to do with
> it and I imagine the developer's tried all sorts of tricks to speed things up.
>
> If the problem only seriously affects network drives, it's likely that the code's
> doing too many calls to query file existence and attributes, all of which will
> suffer on a (slower) network connection.
>
> Be interesting knowing whether other overlay handlers such as TortoiseSVN/CVS
> cause the same problem, whether a simple no-op or one-op implementation
> of _get_state ameliorates things, or whether the added layer of the Python
> interpreter is causing the problems.
>
> As a side note, might be worth tying in to the TortoiseOverlays projects
> which provides for a shared set of overlays for all Tortoise.. projects,
> preventing problems with the rather low system limit. (Don't imagine it
> would help speed things up, tho', since it introduces *another* layer
> into the stack).
>
> Perusing the code at:
>
> http://tortoisesvn.tigris.org/svn/tortoisesvn/trunk/src/TortoiseShell/IconOverlay.cpp
>
> was very instructive! Worth checking the repos out to look at the log, too.
>
> TJG
> _______________________________________________
> python-win32 mailing list
> python-win32 at python.org
> http://mail.python.org/mailman/listinfo/python-win32
>
>
>   
Andrea: Thanks for testing and reporting

Tim: Thanks for giving so much detailed insights to it and of course I'm 
looking forward to read your cookbook about shell extensions written in 
Python ;-). By the way I have nothing to do with TortoiseHg, it was just 
one codeexample discussed on this list last january that I linked to on 
the wxpython list. While Rogers code seems to be a bit shorter, my quick 
and inexperienced look at the codes made me thinking that both examples 
do basically the same. That was the reason for my question. But 
nevertheless either the shell extensions and TortoiseHg are really 
interesting topics. And maybe the TortoiseHg project can also get their 
benefit out of this discussion.

Jürgen