Re: [Pan-devel] Re: GTK/Win assertion

[Gisle Vanem]

"Duncan" <address@hidden> said:

> Read the "Want to fix memory consumption issues" thread.  The old GTK+
> build PAN was using is depreciated.  The author is no longer developing
> it, but recommends switching to a GTK+2.4 build by Tor, which is confirmed
> to work.

Okay, thanks. I've switched to Glib 2.4.2 and GTK+ 2.4.3 which seems to
work nice. I has native Win-thread support AFAICS.

> Another possibility, if it comes to that, would be to implement just a
> (faux) threading system on MSWormOS, using the timer functionality you
> mentioned, so one isn't messing with PAN or even all of GTK+, but just
> implementing the one set of functionality, to be used by anything needing
> it, including PAN.  Maybe that's how it was handled in the new GTK+
> version?

I tried Pan on Win-XP approx year ago. But it ran like melasse. Ethereal on 
the other hand runs smooth (built that myself). So I assume GTK+ has been 
improved greatly the last year. Hence the reason I'd like to build pan myself. 
How did pan on Windows run w/o GTK/Glib thread support? Are there > 1 
branches of pan?

Anyway it looks cool. Only problem I see now is that pan linked as a
GUI app doesn't get a console. So "pan --debug" doesn't do anything. Should
maybe allocate a console when needed. I see Glib/GTK prints to console 
on warnings etc.

--gv