Re: [wmii] snap: 20060309

On Thu, Mar 09, 2006 at 07:58:54PM +0100, Uriel wrote:
> The rules file should be inside wmii, for the same reason the plumber
> rules are inside the plumber: it is simpler and cleaner. They specify

You completely miss the point that the plumber is used by
various applications/9P servers following the philosophy doing
one job and doing it well. wmii is not the plumber, like acme is
not the plumber, but you can easily combine both with the
plumber.

> the default set of tags, and the one that initializes the tags is the
> wm, so the information about how to initialize tags should be stored
> and maintained by the wm because it is the one that knows best the
> context of the creation of the window.

There is no difference, if the wm initializes the tags of a
window after receiving a response to some event OR if it matches
some rules with some YAREL to initialize some tags. In both
cases the decision has been supplied by the user (wmiirc) and
not guessed by the WM.

> Of course once more you are blinded by your total incapability to
> think in more than one single straight line at a time, and assume that
> just because the wm sets the default tags like that, it should not
> allow external programs to set the
> tags, which obviously is a totally senseless conclusion.

Sorry, I never told something like that.

> If you don't learn to step back, and to see how the small pieces fit
> into the big picture, and how all the different design factors
> interact; and instead keep dogmatically following totally arbitrary
> and senseless metrics to measure how things should be done, we are

Even if you know, that you're wrong you insist your proposal
like a an only-child.

> hopeless. Programming is an art, not an exact science, and it requires
> a sense of aesthetics, style and perspective; and like ken said, one
> has to see how the small pieces fit into the big structures.

Yes, exactly. And you propose doing big structures instead of
small pieces which work together in a simple way. Inventing YAREL
to write some tags to some structs is really insane.

> Now go ahead and ignore everything I say and build your extremely
> complex system to push tag initialization logic into external
> programs, that then call xprop, parse xprop output, then write to the
> tag file, and then say they are done figuring out what tags should be
> selected, etc. Great idea, push all the tedious work to the user, to
> save a handful of hypothetical lines of code. Blah, if you only could
> read The Practice of Programming, which has a whole chapter dedicated
> to the power of small specialized languages.

A handful? You mean awk is a handful of lines?

> After taking a look at xprop, I am in awe as to what sort of perverted
> mind conceived such abomination, and how anyone could ever use a
> system as retarded as X; but still, I can't see how we could not just
> pick a handful of attributes (most of which we probably are needed by
> wmii internally anyway) and export those as strings, I certainly see
> no sense in allowing to set any of them, but if we are going to fetch
> them anyway, why not provide a sane interface to them rather than the
> xprop insanity.

Simply because the wm normally only cares for WM_NAME,
WM_PROTOCOLS and WM_HINTS. And these attributes are exported in
the fs. WM_CLASS is somewhat uninteresting for most purposes.
But for tagging they might be most useful compared to all others.

And don't forget, it was not me who proposed using xprop, it was
you. Weak memory recently?

> I never have seen any need for /def/tag, and still think it should not exist.

Uh, that shows that you never think very deep. Which tag should a
window get if no ws exists yet, no other window existed, no wmiirc
already finished its job?

You ever started wmii if there are existing windows?

Regards,

-- 
 Anselm R. Garbe  ><><  www.ebrag.de  ><><  GPG key: 0D73F361