On Monday 23 September 2002 2:46 pm, Kuba Ober wrote:
> On poniedziałek 23 wrzesień 2002 03:49 am, Juergen Vigna wrote:
> > Moritz Moeller-Herrmann wrote:
> > > I am a user, I can't code any C or C++. Sorry. If my bug
> > > reports and criticism are not welcome, tell me so and I
> > > will shut up.
> >
> > You know that this is not so. As John pointed out (well he
> > was a bit grumpy but probaly the pub did close again too
> > early for him to go to bed ;) we cannot fix all bugs (not
> > enough manpower at the time) and WONTFIX doesn't say we did
> > not agree it's a bug it just says this IS a bug but we won't
> > fix it in any known timespan.
>
> Can't the pipes be just unlinked at start of lyx? Cmon, it's
> two lines of code...
>
> If the pipes are in use, the are open by at least one side
> (writing or reading), so unlink doesn't really destroy them.
> They just become inaccessible to everybody but the ones who
> have them already open. Or is it that those semantics don't
> apply to pipes???
>
> Cheers, Kuba Ober
>From my observations when playing with these things before
dissappearing for a week, that's exactly what happens. So I too
can't see any problems. Perhaps, however, this is a
Linux-specific behaviour. Maybe other unices behave differently?
However, the test exists so that the pipe isn't deleted should
there be two or more instances of LyX running at the same time.
There is no clean way of coping with both sides of the argument
other than giving the pipes unique names.
Eg lyxpipe-<PID>.in/out.
Of course if we can test whether the pipe is open before
attempting to unlink it, then we could definitely delete pipes
that are closed...
Angus (ignorance showing once again).
