On Mon, 12 Apr 2010 at  0:19:32 +0200, Tamas TEVESZ wrote:
> 
> looking into how to integrate the shiny new wmgenmenu into 
> wmaker.inst, i came to the conclusion that doing what debian does 
> is a better way.
>
> in short, in debian, "wmaker" is a shell script similar in purpose to 
> wmaker.inst, as in it does the user-level setup *if needed*, then it 
> just execs the real wmaker.
>
> i think we should adopt this. it more or less completely replaces 
> wmaker.inst, except wmaker.inst will modify your 
> .xinitrc/.Xsession/something else too. 

I won't disagree with your judgement on this matter, but on a first look
of debian/wmaker.sh (that's what you mean, right?) I didn't quite became
a fan of what I saw. It's a 70+ line shell script which ultimately for
99.99% of the time will simply execute the last line.

I much rather prefer doing the install stuff once at wmaker.inst, and then 
the other 99.99% of the time simply do a "run the wmaker binary as fast as you 
can" 
and saving every 0.1 sec you can get.

I exaggerate a bit, but I don't like the overhead of loading unnecessary stuff 
at startup
at a theoretical level (ie even if it means a 0.01 sec delay on a fast 
computer).
 
 [ And now I have an Intel SSD and those kind of things become 
less-then-milliseconds issues,
   but until recently I had the worst 4200rpm drive ever made in the history of 
the hardware 
   industry, and loading even a few libraries would mean a stupid delay. Just 
yesterday
   I spent some time to make my wmaker not load the libtiff and libgif and 
thought about
   stripping even more stuff, but strictly speaking I don't need to do these 
kind of 
   things anymore and do it just by force of habit and for the 4200rpm trauma 
:-)]

So if you can find a solution where we don't use a shell script to check things 
at every load, I would be theoretically happier. But if you don't I can live 
with 
sub millisecs delays, I guess... :-)


-- 
To unsubscribe, send mail to [email protected].
  • wmaker.inst Tamas TEVESZ
    • Re: wmaker.inst Carlos R. Mafra

Reply via email to