CLIM mail archive
[Prev][Next][Index][Thread]

Resource management in CLIM

> Date: Fri, 12 Nov 1993 10:57 -0500
> From: Scott McKay <SWM@STONY-BROOK.SCRC.Symbolics.COM>
> 
>     Date: Fri, 12 Nov 1993 09:46 EST
>     From: John Collins <jecollins@mmm.com>
> 
>     Franz Allegro 4.2/CLIM 2.0 Beta2.0
> 
>     We are having serious problems using a CLIM application on a PC
>     X-server. It consistently dies after an hour or so of continuous use.
>     At first we assumed the problem was with the X server, and have been
>     working intensively with the vendor. However, we have been unable to
>     replicate the problem with other (non-CLIM) applications. So, we tried
>     it with one of the CLIM demos, the bitmap editor. We can make the X
>     server crash just by repeatedly bring up CLIM dialogs. Apparently
>     (this is just a wild guess) the other applications are re-using the
>     dialog resources and CLIM is creating new structure every time, and
>     the X server eventually chokes. Is this true? Is there any way to
>     control this? If not, it's hard to see how CLIM can be used as a
>     production tool.
> 
> Presumably it is a bug in the Allegro CLIM back-end that it is not
> deallocating gadget resources.  It hardly seems productive to include
> comments like the last sentence, since it implies that they would
> knowingly create a useless product; the only result is that you will
> just piss them off, which I am sure won't motivate them in any way.

I'm very sorry for that implication, and I apologize to CER and the
folks ar Franz. I don't mean to offend, certainly. It's sometimes hard
to keep your cool when you're being beaten up by users and management.

I guess I've always thought of CLIM primarily as a prototyping tool,
and it's good for that. Also, most Lisp applications don't run
continuously; ours needs to run continuously, day and night, and will
be used all day every day, with users logging in and out while the
system continues to run. We have found numerous weaknesses in CLIM
which would simply not be a problem in a system that you ran for a
while and then quit. It is quite possible that many of these
weaknesses are documentation or implementation problems, rather than
problems in the design of CLIM, but it is very hard for us to tell the
difference. For example, find-port doesn't do the right thing if you
log in from a remote server, log off, restart the server, and then try
to log in again. It just sends windows off into the ether without
complaining.

I suppose perhaps we made a mistake expecting to get a system as
demanding as this working on beta software, but we were strapped for
resources and thought this approach could save us quite a bit of time.

Anyway, I'm sorry if I jumped to an unjustified conclusion and
offended. The fact remains that we are kind of stuck at the moment.
------------------------------------------------------------------------
     John Collins			Phone: +1 (612) 736 0778
     3M Company				FAX:   +1 (612) 733 2165
     3M Center, Building 260-6A-08	Internet: jecollins@mmm.com
     St. Paul, MN  55144-1000		   or     jcollins@cs.umn.edu
------------------------------------------------------------------------
Follow-Ups:
Main Index | Thread Index