IRC channel logs

2013-10-07.log

back to list of logs

***Guest12045 is now known as bananagram
<civodul>Hello Guilers!
<wingo>moin
<civodul>wingo: there's a thread on web review tools on g-p-d
<civodul>would be interesting to share your experience
<dsmith>Morning Greetings. Guilers
<ArneBab>davexunit: ouch - do you have an idea, why the patch is corrupt?
<ArneBab>davexunit: dt varying from frame to frame is exactly what I need: The real system cannot guarantee, that I will get 60 frames per second, but I as programmer have to guarantee, that my physical simulation stays valid (else something on screen suddenly speeds up and slows down again)
<ArneBab>davexunit: yepp, my diffs are from hg, so you might need to use patch -p1 < file.diff to apply them.
<nalaginrut>ni guys
<nalaginrut>hi
<nalaginrut>I'm playing a simple MUD server with modifying our inner server, which is very cool to hack
<nalaginrut>but I found a problem, the inner server always call sanitize-request which is obvious for HTTP like request
<nalaginrut>sometimes we don't need HTTP like request headers, I don't think it should appear in a generic server framework
<ijp>hmm, it would be weird if that happened when you were using a non http server
*ijp goes look
<wingo>it's not really a generic server tho
<wingo>it's a web server
<nalaginrut>is it? I thought server/http.scm is web server
<ijp>well the design implies a use of requests and responses
<nalaginrut>anyway, it's very easy to define new server, if get rid of that HTTP request sanitize
<ijp>the whole module does seem to make a lot of http assumptions
<nalaginrut>I do defined request and response, it's compatible with the inner server interface
<wingo>if you want a generic server thing you will probably be happier making your own server module
<ijp>on a related point, I thought nalaginrut already did that :)
<nalaginrut>yes, I just tried inner server for generic purpose ;-P
<nalaginrut>now that I wrote some code for inner server, maybe I should copy server.scm then modify it
<nalaginrut>alright, I ignored this: "(web server) is a generic web server interface" ;-)
<ijp>well, it's not a lie, per se, if you take web=http
<nalaginrut>but it looks like a generic one ;-)
<nalaginrut>anyway, I can easily modify my own, and I've done it
***sethalves is now known as sethAway
<ijp>protip: adding a couple pks to peval is a great way to grow your compiler output
<wingo>:)
<wingo>there is some built-in debugging support; see the top of the file
*civodul sees monads everywhere
<civodul>the counter and effort thing: a monad!
<ijp>hmm, weird
<ijp>wingo: the problem there is that then it logs too much!
<ijp>now most people will say there is no such thing, but I know better! :)
<wingo>ijp: i think you can set particular categories to log
<wingo>that might be too much for you tho :)
<ijp>this is a very curious bug
<ijp>I'm looking at the patch that introduces it and it does 2 things
<ijp>1) it moves a definition without changing it, and 2) it introduces a loop
<ijp>now 2) isn't causing it, because I checked, but then that leaves 1 :/
<ijp>*magic*
<ijp>ah, hold the phone
<ijp>aha!
<ijp>nothing like making a bold statement about what a piece of code does on irc to make you realise you were wrong
<wingo>especially about peval :)
<ijp>so basically the problem is a call to find-definition did what it was supposed to do
<ijp>but find-definition shouldn't have been called in the first place, because the variable was mutable
<mark_weaver>ijp: good hunting! :)
<ijp>I was just finishing off the mail to you :)
<mark_weaver>do you see how to fix it? (I'm busy with other things now :)
<ijp>I'm not entirely sure yet
<mark_weaver>presumably there must be something in there to keep track of what variables are mutable. otherwise, peval has big problems :)
<mark_weaver>(and more generally, whether an expression might reference any mutable variables or have any effects)
<ijp>well there is, but right now, I'm not sure how to use it
<mark_weaver>ijp: well, a while back you suggested that I use the phrase "I bet you can't ..." more often. so: I bet you can't fix this bug! :-P
<mark_weaver>(another Guile hacker that understands peval would be a very good thing :)
*wingo_ currently working on an rtl compilation bug
*wingo_ breaks down and actually builds a dominator tree
<stis>evening guilers!
<stis>nice hunt ijp:
<wingo_>evening stis :)
<stis>heya wingo_: As always, really nice article on wingolog :-)
<wingo_>tx :-)
***wingo_ is now known as wingo
<wingo>been a little while
<wingo>i have a number of guile things queued up
<stis>yeah, but it feels like a final rtl compilation iog guile is closing in.
<wingo>yes it seems close
<wingo>though when i see bugs like the one i am seeing right now, i wonder ;)
<wingo>the compiler doesn't compile loops correctly, which is pretty basic :P
<stis>ouch!
***sethAway is now known as seth_
***seth_ is now known as sethalves
*tupi writes something so that our today irc chats does not end with 'ouch!' :)
***weinholt` is now known as weinholt
***ijp is now known as prometheus
***prometheus is now known as ijp