pan-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Pan-devel] crashes on OSX :( (Re: Progress update on multi-threading of


From: SciFi
Subject: [Pan-devel] crashes on OSX :( (Re: Progress update on multi-threading of pan uudecode code)
Date: Wed, 21 Mar 2007 00:43:29 +0000 (UTC)
User-agent: Pan/0.125 (Potzrebie)

Hi,

On Tue, 20 Mar 2007 05:56:53 -0400, Calin A. Culianu wrote:
> [...]
> Anyone want to test it?

The patch fits fine here src-wise, but I had to manually tweak the
Makefile.in files (based on what is changed in their related
Makefile.am) since the latest official GNU tools (e.g.
Automake-1.10 &c.) won't produce a working build system on OSX
here as I reported in an earlier post on this list (see "problems
building pan2 from svn" on Feb.27).

The src with your patch then seems to compile fine.  But starting
pan from xterm gives us a crash apparently inside Apple's
libstdc++ (null pointer somewhere) from the git-go.

I've saved a crashreport which includes a backtrace of sorts,
should be small enough just to attach to this e-mail if gmane will
allow.

I re-installed the svn r205 build prior to the patch, and
obviously it is working (using it to post this msg ;) ).

Let me know if I can help further, please, with detailed
instructions?

As an aside, while I have your eyeballs here ;) , since we're
talking about what blocks the other parts of Pan while running,
I'm finding the worst problems are when fetching new headers (more
precisely when they are saved/merged with previously fetched
headers), marking groups read, selecting a group then threading
the headers for display, those kinds of things are blocking say
the downloading queue to the point that the network isn't being
utilised (I have the freeware MenuMeters showing the activity,
which runs as a native OSX task by itself).

Also, another MenuMeters display is showing that only one of these
Dual G5 CPUs is being utilised as well.  Is there some way to
assign threads to "the other one" so both CPUs can be more
balanced at handling work?

fwiw I'm quite certain Darwin/OSX uses a pthreads model, if that
affects development of your code somehow.  I was sure if any such
configure options were available for glib, pango, cairo, and so on
down those trees, I did enable/compile them with pthreads support.

Whatever the case, thank you very much for taking time to do this.
I'm quite interested since we don't have much truly
multi-threaded news-readers for OSX it seems.  :)

(hmmm I think we need to add a platform/version string to the
User-Agent line, to show what flavour of pan was used, eh?)

(on 2nd thot, I don't see how to attach a file to this post,
it might be better to send it to the bugtracker, maybe?)






reply via email to

[Prev in Thread] Current Thread [Next in Thread]