Re: [Evolution-hackers] RH 7.3 filter / threading bug

[Michael Meeks <michael ximian com>]

Hi Jeff / Michael,

On Thu, 2003-11-27 at 15:26, Not Zed wrote:
> Cool.  I didn't think the thread stuff used sockets, but i never
> looked into it too closely.  That code looked like 'standard'
> fork/exec code to me, but maybe its a hangover from much older unix
> days (i.e. sunos4).

	Yes indeed - interestingly stracing the shell shows it doing a close
flood before fork - I suspect this happens to work only because it's a
single-threaded application without the worker thread + pipe - which is
faily gross ;-)

> > 	A quick grep for _SC_OPEN_MAX reveals that the same problem
> > code is cut and pasted [ URGH - why ? ] in HEAD and stable in a
> > frightening number
> 
> I really hate pointing the finger like this, but, blame Jeff on this
> one I think :)

	Is this bug in Jeff fixed yet [ or has it been duplicated in several
places ] ? :-) I'm sure Jeff writes really nice code, but if we cut and
paste more than 3 lines of (even nice) code we're creating a nightmare
for everyone, slowing / bloating the run-time, and avoiding creating a
simpler, more readable, more powerful, re-factored code-base; but
everyone knew that already I guess.

> > 	Anyhow, it was an 'interesting' bug to have to untangle (with Dan's
> > help) :-) the thread helper thing really sucks; and I guess none of use
> > have RH 7.3 systems handy to test on these days.
> 
> But this would mean that they were always having this issue, not just
> with > 1.4.5 as they claimed :)

	That's true, of course they claim to have been having long-term pain
with threading that's much broader than this issue - which would figure
I suppose ;-) anyway, the sooner we can get the quick fix out the better
I imagine; it'd be nice to have it re-factored too in the longer term.

	Thanks,

		Michael.

-- 
 michael ximian com  <><, Pseudo Engineer, itinerant idiot