Re: [Pan-devel] some patches

[Haran Shivanan]

Module 'Charles Kerr' core dumped on Mon, 6 Jan 2003 with the following output

> 
> The rest of the code is fine, though the detail regarding gtk's tag fields
> makes me wonder if there's not a gtk API call to do this already?

I know, it looks like a nasty hack.
At first I was explicitly removing each tag, and then re-entering them
by calling the setup_text_buffer_tags() but for some reason, this is
wrong.
The program becomes unstable and crashes sporadically. Sometimes,
assertions also fail. I got the current implementation from the gtk-devel
archives. I've gone through the gtk_text_buffer api but havn't found
anything that does it cleanly.

> 
> Regarding \n--\n, I think we should accept it for reading as long as we
> are reasonably sure it's at the end of the article.
> Perhaps if the number of lines underneath \n--\n exceeds N (six?)
> then we wouldn't treat it as a signature delimiter.

Or we could make sure it is the the last "\n--" bit in the article.

> 
> message-window.c has some code to find the signature delimiter.
> Haran, would you extract that out and promote that up to util.[ch]
> so that text.c and message-window.c can share it?

Sure. I got confused at first, because I thought the signature delimitor
code was exclusively in message-window.c. Yet my modifications to it
didn't seem to have any effect. Only later I found the actual code that
was removing the sigs was in text.c. So yeah, the sig recognizer really
should be centralized. I'll get on it. 

Haran

-- 
An engineer is someone who will spend three hours
figuring out how to do a two-hour job in one hour