Re: [wmii] Preparation for 3.7 Release.

From: sqweek <>
Date: Thu, 29 May 2008 17:15:29 +0800

 Just been having a look [finally] through the rc.wmii changes from
3.6 to 20080520. Some comments:

* Event-Notice: nice idea, dicey implementation... I implemented
something similar as a seperate action awhile back, that would
remember the last 10 messages and cycle on mouse clicks. Never got
around to sharing (or using) it.
* LBarMenu-3-Delete: Guess what function got unbound as soon as I noticed it? :P
  The generalised menu interface is kind of nice though.
* /tags/1 exists even with no clients now?
* # For the time being, this file follows the lisp bracing convention.
  This comment totally failed to surprise me; after debugging a couple
of things and absorbing the scripts, the *code* started reminding me
of lisp. I half suspect you'll take that as a compliment, but it got
me wondering... Why are we using rc? rc's forte is environment
manipulation and process spawning, not meta programming and api
construction (ok, rc.wmii does its fair share of the former aswell).
In fact, considering the interface now provided by wmii.rc, I wonder
why we use 9p at all. wmii could be using any sort of RPC protocol
behind wi_foo and no one would know the difference. The greatest one
for me was:

fn wi_readevent {
        wmiir read /event

 What. The. Hell. wi_readevent is referenced exactly once, for those wondering.
 I suspect this is half the point - hide details of the 9p interface
behind a "nice" rc interface. But isn't that the exact point of 9p? To
hide the details of X11 and window management behind a "nice" 9p
interface? The mere existance of wmii.rc just screams at me "Why, WHY?
Why do we need this extra level of indirection?" The obvious
conclusion is that our current 9p interface sucks, but I wonder if
there aren't other factors...

Received on Thu May 29 2008 - 11:15:34 UTC

This archive was generated by hypermail 2.2.0 : Sun Jul 13 2008 - 16:43:05 UTC