freenode/#clim - IRC Chatlog
Search
8:11:19
jackdaniel
I was referring to the fact, that where you have presentation-type in publish-selection, it may be a list of presentation types
8:13:47
jackdaniel
selection protocol is orthogonal to presentations, so I don't see much controversy with it
8:15:18
loke
I'm just trying to think of cases where you could have more than one possible presentation type for a given object.
8:18:02
jackdaniel
I have given an example with abstract door/wood types. more practical one: you publish a rectangle which is a "documentation figure", "output-record" and "polygon"
8:20:36
jackdaniel
but objects and presentation types are very disjoint, you may say: (with-output-as-presentation (stream 3 :type 'string) (draw-rectangle …))
8:21:16
jackdaniel
note that 3 is not a string, but you will be able to select this rectangle if input context looks for a string (and it will blow later if you really expect string in there)
8:27:10
loke
But would you pass all of those types to the publish function? Wouldn't you want to publish as a given type, only?
8:27:10
loke
But then again, I'm pragmatic, and since it doesn't _remove_ functionality, I have no real issue with it.
8:50:47
jackdaniel
no for three reasons: 1. it needs to be remade to have clean commit history (right now I've just slapped a few commits on top to show what I have in mind), 2. I'm still thinking about synchronous api, 3. documentation is incomplete
8:51:28
jackdaniel
also, nomen omen, nobody has reviewed my code so far and I'm sure I've planted numerous issues in there
8:59:47
jackdaniel
also could anyone try to reason with me about process-next-event? (http://bauhh.dyndns.org:8000/clim-spec/8-1.html#_303)
9:27:02
jackdaniel
beach: maybe you will be interested in this non-unix operating system comparison to unix: https://bzdww.com/article/163937/ (fuchsia os)
9:28:02
jackdaniel
of course there is a lot of FUD in there, but some interesting insights and ideas too
9:49:32
loke
Based ona quick reading for this, it seems that the permission model is still centred around the concept of a process.
9:55:31
jackdaniel
I'm going to get rid of a non-standard protocol function get-next-event (which is just trampolined from process-next-event)
10:06:56
jackdaniel
there is some return value mangling and process-next-event calls distribute-event on a result event, but yes, get-next-event is just a clutter
18:18:13
jackdaniel
based on a name and prototype, what would you expect this function to do? (it doesn't exist) -- (defgeneric event-listen-or-wait (sheet &key timeout wait-function))