tips for Opera users

[Jim Kronebusch]

Thanks for all your testing Gavin.  I could compile granparadiso last night but it
didn't seem to apply the patches.  If you would want to post a howto with applying the
patch that would be great, then I could play with those numbers as you suggest.

Non-Gecko based browsers do not cache the pixmaps at all.  Opera is actually faster than
Firefox in displaying that page and does not put any load on the client.  This makes me
think that there needs to be a way to disable it completely.

On Tue, 18 Sep 2007 10:57:03 +0100, Gavin McCullagh wrote
> Hi,
> 
> I got the patch to apply and work.  I can't say for sure that it will fix
> your issue but it might at least alleviate it.
> 
> If I understand correctly, Federico's patch doesn't change the fact that
> firefox has to load all of the images, as pixmaps, into video ram to
> display the page.  I presume all browsers must at some stage load each
> image in that way.  It seems his patch causes the pixmaps to be removed
> after a short period, unless the particular picture is actually on-screen
> at that moment.
> 
> 	http://primates.ximian.com/~federico/news-2007-09.html
> 
> This seems to work well enough.  However, if you try to quickly scroll
> through the screen quickly, you'll see the video RAM spike to similar
> values as before as every image has to be painted before any get removed.
> For this reason, I think it'll still be possible to crash the thin client,
> though perhaps it'll be less likely, as firefox will be purging memory.
> 
> One small part of the patch is:
> 
> +static int
> +get_discard_timer_ms (void)
> +{
> +  /* FIXME: don't hardcode this */
> +  return 5000; /* 5 seconds */
> +}
> 
> which suggests the time limit is 5 seconds.  We could perhaps try lowering
> this to make the cache time shorter and make the crash less likely.
> 
> On Tue, 18 Sep 2007, Gavin McCullagh wrote:
> 
> > > http://www.carteretcountyschools.org/bms/teacherwebs/sdavenport/artgallery6.htm)
> 
> > res-base Wins  GCs Fnts Pxms Misc   Pxm mem  Other   Total   PID Identifier   
> > 6400000   276  477    1  381  608   294074K     32K 294107K 21296 Mrs.  Davenport's
Web - Gran Paradiso
> 
> At different moments with that webpage loaded, I now get:
> 
> res-base Wins  GCs Fnts Pxms Misc   Pxm mem  Other   Total   PID Identifier   
> 4c00000   110   99    1   48   88   281122K      7K 281130K 23604 Mrs. 
> Davenport's Web - Gran Paradiso 4c00000   110   80    1   29   69    66303K    
>   7K  66310K 23604 Mrs. Davenport's Web - Gran Paradiso 4c00000   139  125   
>  1   57  144      550K     10K    561K 23604 Mrs.  Davenport's Web - Gran Paradiso
> 
> the bottom one is where firefox is on a different virtual desktop and is
> therefore not being painted at all.  The top is where I'm scrolling through
> the page and the middle is just sitting on the page.
> 
> Firefox is very slow at rendering this page now.  Every time I go back to
> that desktop, there's a delay in rendering it.  However, perhaps that's not
> as bad as a crash.  The machine I'm on is a dual core 64-bit athlon with
> some sort of flashy ATI video card, but I am using the vesa driver (to
> avoid proprietary video driver), which is not the quickest thing.
> 
> Gavin
> 
> -- 
> edubuntu-users mailing list
> edubuntu-users at lists.ubuntu.com
> Modify settings or unsubscribe at: 
> https://lists.ubuntu.com/mailman/listinfo/edubuntu-users
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by the Cotter Technology 
> Department, and is believed to be clean.


Jim Kronebusch
Cotter Tech Department
453-5188


-- 
This message has been scanned for viruses and
dangerous content by the Cotter Technology 
Department, and is believed to be clean.