Re: [dev] PATCH for/back navigation with xprop

On Fri, 3 Jun 2011 23:16:49 +0000
Bjartur Thorlacius <svartman95_AT_gmail.com> wrote:

> It's just that I believe spawning processes and creating windows
> should be dirt cheap operations. Optimally, almost nothing should be
> done in those functions.

With Gtk+ 2.OMGWANTMOARBLOAT invading computers everywhere, your belief is most understandable! Seriously though, adding a single extra string match isn't going to make much difference, nothing like whatever Gtk+ 2 does. I wouldn't be surprised if the entire window-matching list of a large FVWM config fit in 1 page of virtual memory, and the window manager as a whole was designed to use little memory. For a long time it used much less than TWM. As I said it ran smoothly on an 8MB 486.

I'm a little bit.. scared of perfectly decent things being cut out in the name of sucklessness to be honest. Years ago I had the most terribly fancy themes for Gtk+ 1.x on a 333MHz AMD K6/2. Scaled textures absolutely everywhere on a crap CPU with unaccelerated graphics came at a barely noticeable cost in startup time; nothing I could detect without a back-to-back comparison. What I'm trying to say is you can do extremely crazy stuff without adding a noticeable delay to creating a window.

I don't know what Gtk+ 2 does to make my netbook swap for 3 full seconds before opening a menu. Whatever it does must stem from some seriously diseased design.
Received on Tue Jun 07 2011 - 00:28:14 CEST