[Thread Prev][Thread Next]   >Date Index >Thread Index

Re: [wmx] sluggish netscape with channels

Damion Yates - Fri Jan 28 12:33:02 2000

On Wed, 26 Jan 2000, Joel Reicher wrote:

> Damion Yates wrote:

[slightly edited]
> > However, when I eventually came to compile and run wmx (6pre1), I
> > immediately found sluggishness on the 2nd channel in netscape, I first
> > noticed to my surprise how scrolling down with the scrolly wheel of my
> > mouse was really slow, the X resources were set to do 4 lines at a time,
> > and I saw each line movement on down, but it was instant on the up move.
> > Then I simply used pageup/down and found weird slowness, just like I get
> > at work.  This was just on slashdot.org as well so not such a large page.
> > After it had scrolled down and up a few times to the bottom, it sped up,
> > even if I switched to channel1 and back to 2 again, it was fast.  I went
> > to another page news.bbc.co.uk and found the same.  I tested other
> > virtual windows on olvwm and enlightenment and was unable to replicate it
> > this time, it seems there is something up specifically with wmx.

> FWIW, I experienced this as well and I think it's related to the focus
> problem I mentioned a few weeks ago, because if focus was brought back to
> the netscape window `explicitly' by clicking on wmx's frame for the window,
> the problem went away.

Aha, yes this does help.  I've found that all I need to do is bring up a menu
on the root window to switch focus away for a mo, then come back and the
netscape is cured.  I'd at first experimented with a ~/.wmx/fix\ netscape
script that spawned xterm -e pause, where pause just starts and exits.  I had
the geometry of the xterm large so it would always give me time to move my
mouse to it, therefore changing focus properly.  Then it would die and netscape would
gain focus, and the side bars etc.

This all wasn't necessary as just the envoking of a menu changes something.
So I suspect a nasty kludge would be to simulate all but the actual drawing
of the root menu on a channel switch thus running through which ever wmx code
is causing the fault to go.  I'll try out some ideas, but it'll be really
lame :) better somebody gets in there first, I'm an asm/perl coder these days
(odd combo I know), my C is rusty, C++ dire.


Damion Yates - Senior Internet Operations Engineer - Internet Services
email: Damion.Yates@bbc.co.uk - phone: +44 1737 839510