Search
Wednesday, 21st of June 2017, 1:25:47 UTC
1:27:14
Zhivago
Largely for convenience -- scheme doesn't do that, for example.
1:27:51
Zhivago
nil isn't the same as (nil) for many operations, but it is the same for (car '(nil)), yes.
1:29:04
fvs
okay, thanks. so then does scheme throw an error on (car nil)?
1:31:12
fvs
because that's what haskell does (or rather, that's what it does on (head []), which seems to be equivalent)
1:37:18
PuercoPop
fvs: because convenience and correctness sometimes are at odds. Scheme chose the latter
1:37:19
minion
PuercoPop, memo from flavio81: Estas listo para conocer anuestro salvador, Common Lisp?
1:44:00
fvs
PuercoPop: Ok, thanks, that makes sense
2:40:55
beach
Good morning everyone!
4:02:09
fiddlerwoaroof
Does anyone here know how long Lispworks takes to respond to emails?
4:45:15
vtomole
I'm getting a "There is no applicable method for the generic function" Even thought i have an accessor to that slot value..
4:54:08
beach
vtomole: Paste some code?
4:56:15
Zhivago
vtomole: Perhaps you're calling it with arguments of a class that you have not specialized it for?
5:00:05
pillton
beach: It seems weird to wrap everything a compilation unit. All of your macros would required eval-when.
5:00:30
pillton
Sorry, everything in a compilation unit.
5:00:54
vtomole
beach: Trying to run some very old program:http://faculty.hampshire.edu/lspector/qgame.lisp
5:02:13
vtomole
http://paste.lisp.org/display/349177
5:03:14
beach
vtomole: The error you show doesn't seem to mention any lack of methods.
5:03:54
slyrus
any sharplispers around?
5:04:57
beach
I have commit privileges.
5:05:02
vtomole
This is what i get:http://paste.lisp.org/display/349178
5:05:11
slyrus
I want to merge my own pull-request on log4cl but figured i
5:05:21
slyrus
'd check to make sure no one objected
5:05:49
beach
vtomole: You are calling it with a list it seems.
5:07:13
beach
vtomole: This: ((#<QUANTUM-SYSTEM {10083D3643}>)) indicates that the argument to the function is a singleton list containing the system, but the accessor is defined for a system and not for a list.
5:07:47
vtomole
So if i take the car of that list..
5:08:09
vtomole
Thanks! I was stuck on that for an hour
5:08:37
beach
vtomole: It pays off to understand error messages. They can be a bit cryptic sometimes, to say the least.
5:08:53
slyrus
beach: ok, you should be good with: https://github.com/sharplispers/log4cl
5:09:06
slyrus
and most recent SBCL and mcclim. if not, I'd like to know about it.
5:09:21
beach
Define "to be good with".
5:09:31
vtomole
Yes. And this code, in my opinion, is not well written (readability-wise)
5:09:53
slyrus
not have compilation errors
5:10:09
vtomole
Even though they use a class which should make things easier. And everything is in one file....
5:10:10
beach
I mean, what am I supposed to do?
5:10:25
vtomole
Im just talking to myself :)
5:10:36
slyrus
use a new sbcl to load mcclim
5:11:15
beach
But what do I do with https://github.com/sharplispers/log4cl? Install it, rather than what is available in Quicklisp? Something else?
5:11:45
slyrus
yes, clone that into ~/quicklisp/local-projects
5:14:33
beach
I REALLY HATE this kind of activity.
5:18:54
slyrus
can you be more specific? Do you hate that mcclim now depends on log4cl? Do you hate that log4cl uses internal symbols from SBCL that aren't likely to be stable? Do you hate that SBCL changed it's use of internal symbols?
5:19:10
beach
I hate doing system administration.
5:19:39
beach
I try very hard to get real work done, rather than catching up with things that change around me.
5:20:09
slyrus
OK, but the alternative (the things around you never changing) is worse
5:20:23
slyrus
gratuitous bad, breaking changes should be avoided
5:20:29
beach
This is the first time I use local-projects. I have not used that method before.
5:20:46
slyrus
Ok. That's where I keep my copy of mcclim.
5:20:52
slyrus
well, a link to it anyway.
5:21:10
beach
I had my Lisp hierarchy before Quicklisp existed, and I didn't want to change it.
5:21:12
slyrus
and now log4cl, until that gets updated in quicklisp
5:21:49
slyrus
ok, fair enough, but I'd argue the benefits to using quicklisp outweigh the costs, even if one has to adopt to some new conventions
5:23:05
beach
Still the same DEFLATE error.
5:23:47
beach
I checked *inferior-lisp* to make sure.
5:23:59
slyrus
what version of retrospectiff?
5:24:08
slyrus
and can you post a stacktrace?
5:24:26
beach
I don't know. How do I check the version of retrospectiff?
5:25:03
slyrus
first: (asdf:component-pathname (asdf:find-system :retrospectiff))
5:25:37
beach
#P"/home/strandh/quicklisp/dists/quicklisp/software/retrospectiff-20170403-git/"
5:26:30
slyrus
and if you do: (ql:quickload 'deflate)
5:27:29
beach
The error is now gone. Did I just update SBCL for no good reason?
5:27:38
slyrus
lots of good reasons :)
5:28:09
slyrus
5 months of impressive bugfixes from dougk et al...
5:28:26
beach
None that I have been suffering from.
5:28:44
slyrus
perhaps there was an update to ASDF in that period that fixed the dependency problem
5:29:27
beach
OK, the application now starts, but as soon as I try to change the zoom level, I get the same error: Vector length (1) doesn't match declared length (4).
5:30:29
beach
Oh, I should pull again?
5:30:47
slyrus
no, that should be fine. can you post a stack trace?
5:34:11
beach
Murii: Very profound.
5:37:44
vtomole
beach: Any non-cleavir SICL related tasks I should look into?
5:39:03
beach
vtomole: Maybe the sequence functions. But consult with me before making changes.
5:39:36
beach
Murii: There is a famous paper about Emacs called "Prose and CONS".
5:39:52
vtomole
beach: I will never make changes before consulting you, Bike taught me that :)
5:40:14
beach
Murii: http://dl.acm.org/citation.cfm?id=802784&preflayout=tabs
5:40:46
beach
vtomole: There is lots of old code, and the code that corresponds to the ELS paper is incomplete.
5:41:37
Guest51760
** NICK jack_rabbit
5:42:27
vtomole
I presume I should start reading the small module first
5:43:10
beach
vtomole: No, that one is obsolete.
5:43:23
beach
vtomole: Read the paper first.
5:44:31
beach
vtomole: The system is paper-sequence.asd
6:06:54
pillton
beach: What I said is probably wrong now that I have re-read with-compilation-unit.
6:07:33
beach
pillton: Yeah, it didn't sound right. I think all it does is to delay warnings.
6:08:08
pillton
beach: It does strike me as odd to want: (with-compilation-unit () (dolist (f files) (load (compile-file f))))
6:09:00
slyrus
ok, I can compile here with debug 3 safety 3
6:10:23
beach
slyrus: It compiles and runs. But when I try one of the sliders I get the error.
6:11:02
slyrus
ah, got it here too :)
6:14:06
flip214
slyrus: I get no error... just works as expected, with a few repaint glitches.
6:14:08
slyrus
Ok, without that declamation, no problem. hmm....
6:14:57
flip214
and the red thing becomes a rectangle...
6:16:30
slyrus
yeah, that should be an oval. I don't think that's my bug :)
6:16:38
beach
slyrus: That's not how I would put it. Without that declamation, the behavior is incorrect without being reported. With it, it is incorrect, but reported as such.
6:17:37
slyrus
beach: fair enough, but I was talking about the oval thing
6:18:00
beach
Ok, without that declamation, no problem. hmm....
6:18:10
slyrus
now to track down the problem...
6:18:58
|3b|`
ACTION interprets with-compilation-unit as delaying the decision to signal the warning rather than just the actual signaling, though i don't see anything actually that specific when i look at how it is specified
6:48:28
slyrus
beach: here's an isolated instance of the problem: http://paste.lisp.org/display/349185
6:49:42
beach
Oh, it has to do with the fill pointer?
6:50:31
slyrus
question is who is at fault here? what's the type of a vector with a fill-pointer?
6:50:39
beach
So the "type-of" v does not match the "type of" v.
6:52:49
specbot
http://www.lispworks.com/reference/HyperSpec/Body/f_tp_of.htm
6:53:24
specbot
http://www.lispworks.com/reference/HyperSpec/Body/f_descri.htm
6:54:15
slyrus
Ok, I've got a fix. my bug, I think.
6:54:37
slyrus
If type-of returns a list, I should only trust the car of the list.
6:55:14
slyrus
or, arguably, SBCL is being overly pedantic here.
6:55:26
slyrus
now I understand why you never want to update your code :)
6:57:07
specbot
http://www.lispworks.com/reference/HyperSpec/Body/f_map.htm
6:57:41
beach
dtornabene: specbot is very handy for transmitting information about the standard to some #lisp participant that you are having a conversation with.
6:59:25
beach
slyrus: Your fix sounds reasonable to me.
6:59:50
slyrus
so, what's the proper way to, e.g., call #'identity on a sequence using map, if one doesn't know the type of sequence but wants to return a sequence of the same type?
7:00:00
slyrus
OK. I'll commit that.
7:00:09
dtornabene
beach: hadn't seen it before, but I mostly lurk here. thanks for the explanation
7:01:38
specbot
http://www.lispworks.com/reference/HyperSpec/Body/f_cp_seq.htm
7:08:30
slyrus
beach: yes, that's an elegant fix. Still, I wonder if this isn't an SBCL bug.
7:16:37
slyrus
ok, so the question is do I call copy-seq on the sequence, or do I only declare the type to be the car of the list, if the type is a list? I'm leaning towards the copy-seq approach.
7:18:12
beach
It seems like a tossup to me.
7:19:11
beach
But that might not matter much.
7:30:06
slyrus
ok, fix pushed to my github repo for McCLIM
8:55:27
jackdaniel
els videos are online: https://www.youtube.com/playlist?list=PLA66mD-6yK8yi-nggbOF1dWusHnu2u6hw
Wednesday, 21st of June 2017, 13:25:47 UTC