freenode/#lisp - IRC Chatlog
Search
14:29:23
lieven
there was a whole cargo cult movement at the time that was adamant about every construction having only one entry point and one exit point
14:33:37
Shinmera
Well, CL does that too to a large part, but in a way that hides it away neatly. Explicit RETURN/-FROMs are pretty rare.
14:35:12
pjb
Shinmera: non-local exits are all over the place in CL. What saves us is unwind-protect.
14:38:53
lieven
the problem generally isn't with the guy who comes up with these insights but with the management who sees it as a silver bullet that will take care of all problems.
14:39:33
epony
in several decades language tools vendors created incompatible products for feature in an attempt to gain market share
14:41:13
epony
the opinionated nature of computer scientists is for a reason quite different than pop culture modern times
14:48:43
beach
Second Climacs now has a gutter and it is used to show incorrect indentation: http://metamodular.com/gutter.png Of course right now, it doesn't know how to compute indentation for DEFMETHOD and LOOP.
15:03:45
pjb
beach: I don't agree (and I suppose slime doesn't either) with the indications given in that gutter.png…
15:04:20
beach
I don't either. Like I said, it doesn't know how to compute indentation for LOOP and DEFMETHOD.
15:06:50
beach
Much better parser, plus I want to plug in Cleavir so that there is program analysis at typing speed.
20:59:00
jackdaniel
Xach: yes, api is literally the same as sbcl (minus stuff added later) - even package nickname is sb-bsd-sockets
21:00:49
jackdaniel
yup, there isn't one on usocket - I wanted to add it and make PR for a long time now, but never found time to get to it
21:01:04
Xach
I guess my question is then "why does usocket not support af_unix/af_local" in the nicest way possible.
21:01:21
Xach
Is it because of OS portability issues? Implementation portability issues? Lack of someone doing it?
21:03:22
jackdaniel
I think there was an issue for that on usocket tracker, but I don't remember for sure
21:07:47
jackdaniel
ACTION directs Xach's attention at dep-* files in portable clx in sharplispers repository
21:10:50
Xach
yes, though comm::*socket_af_unix* makes it seem like there's some support deep inside somehow
21:11:05
Xach
FLI is the Foreign Language Interface. I think that is possibly a nicer name than FFI.
21:12:31
jackdaniel
here you have less 'hacky' version of the same: https://github.com/binghe/lispworks-udp/blob/master/unix.lisp
21:12:49
jackdaniel
I'm getting back to my precious layout protocol off-by-one errors in McCLIM, laters o/
23:01:14
z3t0
Can someone direct me to a well structured project that explains how to properly write a package using ASD and quicklisp and essentially just a guide for how to properly structure a project
23:22:41
krwq
are there any known problems with cffi callbacks when native part calls them inside of the thread?
23:32:25
rme
I know that ccl goes to some trouble to support calling back into lisp from foreign threads. I'm not sure what sbcl does in that case.
23:34:10
Shinmera
It used to require thruption and wtimer features, but I think that's no longer necessary.
23:43:18
krwq
Shinmera: define old - mine is 1.3.1 - so calling back works but it crashes when I try to funcall a lambda in that callback
23:48:45
krwq
Shinmera: do you have any script for building from sources somewhere? last time I tried using the version from sources I was hitting various issues
23:49:27
krwq
my current script is doing: echo '"1.0.99.999"' > version.lisp-expr; ./make.sh --fancy; sudo ./install.sh
23:51:58
Xach
Yeah, if you don't use latest (which I personally find very safe), the latest release tag of the form sbcl-<version> is usually a good bet.