freenode/lisp - IRC Chatlog
Search
15:23:45
Fare
fe[nl]ix, rpg wanted to better formalize asdf:test-system and test-op if your travis-ci initiative for CL leads to bright ideas as to how these could be improved, all the bette
15:23:45
minion
Fare, memo from jackdaniel: I've solved a problem, now a bit of coding and problem should disappear - I doubt tough if it ever worked for systems of the same name
16:02:41
oGMo
also cl-turtle and possibly some others, sadly a bit hard to search for since you can't specify context in any search engines :P
16:17:03
Firedancer
Do anyone of you use Lisp on Arch? Seems like almost all the packages in AUR are old/broken
16:18:38
oGMo
but definitely use quicklisp.. and if getting a lisp _is_ the concern, it's generally best on any dist to build your own
16:20:05
Firedancer
But now Matlisp seem to need CFFI....which need Alexandria....and both packages are old and abandoned in AUR
17:49:14
jasom
If it isn't going to be called "1.0" perhaps Xach should call it "gamma"; not finished, but more stable than a beta
18:39:54
francogrex
has anyone tried the cl-bayesnet. I tried to serialize/deserialize an object but i get this error: http://paste.lisp.org/display/152607
18:40:07
aeth
quicklisp shouldn't be beta, but you also shouldn't just call it "gamma". Skip a few, e.g. 8. That puts you at lambda.
18:40:56
francogrex
same happens if i used another persistence/serialization lib like cl-persistance ... why?
18:46:27
francogrex
(setf (cl-bayesnet::compiled *join-tree*) nil) then store restore => The value NIL is not of type CONS ...
18:47:03
francogrex
i have the possibility to remove the function but it is expecting a cons somehow
18:54:34
jasom
francogrex: hmm I can store/restore a fresh make-instance of net without any problems
19:01:19
jasom
francogrex: I would guess that the nil error isn't from the :compiled, but from something further along (e.g. node-vec)
19:01:40
scymtym_
jasom: globally proclaim increased safety or restrict compiler policy to reproduce?
19:03:45
jasom
scymtym_: there is no type restriction on that slot, so it's almost certainly not where the problem is
19:03:59
jasom
* suspects that if francogrex were to set compiled to 'foo it wouldn't say "FOO is not of type CONS"
19:04:46
pegu
Are there any good s-expr matchers available? I've seen this http://www.defmacro.org/ramblings/s-query.html but I would expect something with wildcards, registers etc.
19:07:00
oGMo
but that may not be whta you're after, since it's a very specific sort of matching, and no registers iirc
19:08:02
francogrex
1. COMPILED: FOO , but still => The value NIL is not of type CONS so it's something else
19:09:13
pegu
oGMo: That looks more what I'm looking for - thanks! I can do without registers if it returns the matched sexpr
19:10:48
oGMo
doubtless, optima as suggested is another entirely different way to do things, and obviously destructuring-bind
19:30:18
eudoxia
drmeister: how are you progressing with Cleavir? i experimented with it over the weekend and built a really tiny AST-to-JS "compiler"
19:32:14
jasom
francogrex: node-order num-nodes and node-vec are all unbound after loading from the dne
19:35:12
jasom
francogrex: I found the issue, the table inside the node named visitasia has dimensionality 0
19:38:58
francogrex
yes not sbcl bc also in ccl yiou get the weirdness VISITASIA NIL #(VISIT NO_VISIT) #0A#(0.01D0 0.99D0)
19:39:43
francogrex
thank god, I was not ready to change a lisp implementation just for that, I am used to sbcl
19:40:16
jasom
francogrex: bug fixes come *fast* once you report them; I've found 2 over the years and both were fixed the same week
19:42:33
Fare
jasom: small enhancement requests with patches, on the other hand, may take years to get committed...
19:43:05
jasom
It seems like every time I ping, they are in the middle of a release so don't want to add new features
19:53:19
jasom
francogrex: ah, probs is supposed to be a list-of-lists but is in fact just a list in parse-netica-node
19:57:10
jasom
francogrex: really not, but bn directly does a (make-array nil :initial-contents ...)
20:04:37
francogrex
ok, but that aside why ccl for example does not complain when restoring such object while sbcl does?
20:11:57
francogrex
nyef_: you probably have not followed the discussion bewteen me and jason. we're trying to understand whether difference between sbcl and ccl is related to a bug
20:13:57
nyef_
Heh. My first serious Lisp project, I ended up using 0-dimensional arrays quite a bit.
20:14:42
XachX
resttime: What library? If you interrupt it, what do you see in the backtrace? What Lisp?
20:18:01
resttime
XachX, in the backtrace there's a lot of [REINITIALIZE-SOURCE-REGISTRY-AND-RETRY] Retry finding system imginu after reinitializing the source-registry.
20:20:37
francogrex
jasom: still though this fixes it when compiled is nil however when it is a closure ccl restores while sbcl still errs...
20:21:42
Oladon
resttime: I could be wrong, but your .asd looks incomplete (missing defpackage/in-package)
20:22:03
francogrex
try (bn:use-join-tree *join-tree*) , (cl-store:store *join-tree* "net") , (cl-store:restore "net")
20:34:51
resttime
oh and (ql:quickload "xmls") also doesn't work on a new instance of sbcl because it complains the package uiop doesn't exist
20:41:20
resttime
i cleared sbcl fasl cache, reinstalled quicklisp, (ql:quickload "uiop"), (ql:quickload "xmls")
20:43:13
resttime
dlowe, yeah though i think i might be able to reproduce the problem with other packages, i'm trying to find out how
20:44:08
White_Flame
digiorgi: generally, no. You'd normally want to output code that is then further macroexpanded
20:44:27
resttime
dlowe, it also used to work for me too, i've been trying to figure out the problem yesterday
20:44:54
White_Flame
manually expanding macros becomes a mess of what context & environment the expansion occurs in, because interaction with the current environment is typically why you want to force expansion to happen right then
20:46:31
White_Flame
digiorgi: another alternative is to convert your inner macroexpansion from a macro to a plain function, then call that function inside your outer macro
20:46:44
francogrex
i understand. I had same issue with a library loading into sbcl... no change and then one day BANG!
21:01:40
francogrex
it's the updates of asdf, the packages and the interactions between sbcl and those
21:03:24
francogrex
maybe quicklisp should send a warning like: your lisp version is too old so don't bother...
21:10:16
pjb
AJavaIdiot: https://www.google.fr/search?q=lambda+calculus+introduction&ie=utf-8&oe=utf-8&gws_rd=cr&ei=J0G5VaTPFMH8UuzXlLAG#q=lambda+calculus+introduction&newwindow=1&safe=off&tbm=bks
21:11:05
AJavaIdiot
I actually wanted recommendations, now just whatever google decided to rank at the top
21:12:19
pjb
AJavaIdiot: You know, since I have enough hard disk space, I resold my books and only kept scans of them and nowadays, with all the material on the web, I don't even bother to look for books, I just google my questions and learn from the web.
21:13:24
AJavaIdiot
pjb: That doesn't work for me. For me to read something on the computer and actually enjoy it, it has to be something really great like land of lisp
21:27:16
francogrex
this seems good enough: http://palmstroem.blogspot.be/2012/05/lambda-calculus-for-absolute-dummies.html just use a printer, bind with a stapler if you are so averse reading offa screen
21:27:50
PuercoPop
minion: memo for rpg: Idk what is normative but if you prepend unix/ to your unix socket file, xlib::get-default-display works as intented.
21:32:39
jasom
francogrex: sbcl is more strict about type declarations than any non-cmucl-derived lisp
22:07:45
pjb
minion: memo for francogrex: that's why this https://www.youtube.com/watch?v=FTSAiF9AHN4 doesn't exist, and this https://www.youtube.com/watch?v=E3keLeMwfHY doesn't exist.
22:29:44
minion
rpg, memo from PuercoPop: Idk what is normative but if you prepend unix/ to your unix socket file, xlib::get-default-display works as intented.
22:36:10
PuercoPop
and that if the protocol is not specified it is a best effort to find the 'fastest' protocol
22:36:44
rpg
I asked the Xquartz people and they don't seem to view the X docs as constituting a spec to which they must comply.
22:36:50
PuercoPop
check http://cgit.freedesktop.org/~chr/libxcb/tree/src/xcb_util.c _xcb_parse_display. It does the same logic that clx does
22:38:24
PuercoPop
of course a more user friendly approach in clx would be to match what it things is the protocol string to an enumeration of valid protocol names
22:54:05
rpg
PuercoPop: one thing that's confusing me as I try to fix things is that CLX uses "host" for a lot of variables, and sometimes it seems that "host or socket" might be ok, and sometimes "no, this should really be a host."
23:02:46
PuercoPop
Idk how to setup Xephyr to reproduce your environment so I can't really help you there.
0:02:02
pjb
resttime: when it will really feel surreal, is when you will run the same program on Linux and on Mac and get the same (or similar) GUI.