freenode/lisp - IRC Chatlog

3:35:37 drmeister ONE-something-something

3:35:38 beach Sounds like it's time to buy a new one.

3:35:55 beach OK, never mind. I'll check the logs.

3:36:06 nyef Sounds like a familiar problem... It's not a Thinkpad, is it?

3:36:17 drmeister This is a one year old macbook pro that Apple gave me to replace my previous one a year ago that also had mysterious, irreproducible video issues.

3:37:18 drmeister Today was terrible. I'm going down to Washington on the train to give an important talk at the Department of Energy (seriously) and my laptop keeps crapping out on me.

3:37:47 drmeister It's a good thing I went down with a colleague - I was able the get the presentation converted to Powerpoint and on to his Lenovo.

3:37:54 Zhivago Tell them that you need Tesla's global wireless power system up and running pronto.

3:38:29 nyef ... Global wireless power system? Isn't that AM Radio?

3:38:41 stassats the sun?

3:39:13 drmeister I was contemplating a drive by the Apple store and throwing it on their doorstep. Then I remembered that I'm born Canadian and we are raised better than that.

3:40:03 Zhivago Nah. He wanted to supercharge the ionosphere and tap into the telluric currents.

3:40:45 drmeister I hate irreproducible hardware problems. Just break already and I can get it fixed.

3:41:21 drmeister I've got it sitting upstairs with the screen black and I'm going to let it run all night. Maybe there's some tin whisker on the motherboard that will fry the thing.

3:41:28 stassats * has a reproducible problem, but i don't know what causes it

3:42:51 drmeister Sorry - no more bellyaching from me.

3:43:04 nyef I at least used to have an intermittent screen-going-blank issue, but I always figured it was software. It hasn't happened since my last full reinstall, so I'm hoping it's gone for now... And I have a new computer coming at some point anyway.

3:43:51 stassats another issue i have is the screen not going blank

3:44:07 nyef (Well, somewhat software, I figure the embedded controller was killing power to the entire video system for some reason.)

3:46:15 beach drmeister: Problem of forward reference fixed.

3:46:28 drmeister beach: I was able to generate ASTs into file 3 of about 70 source files.

3:47:12 beach drmeister: Is that good or bad?

3:47:17 drmeister There were issues with specials not being recognized as such and one case where an (eval-when (eval compile load) ...) wasn't being evaluated.

3:47:53 drmeister That's good - that's a lot of forms.

3:48:00 beach OK. :)

3:48:08 beach We'll get there.

3:48:46 drmeister That's about where the screen when black.

3:49:44 drmeister I'd like to generate some HIR and start converting HIR to LLVM-IR. I'd like to have a taste of how that is going to work.

3:49:58 beach Sure.

3:51:25 drmeister You put a "meister" directory in SICL - I presume that was copied from my fork? I'm trying to wrap my head around how that works with my fork and your original and whether I should keep the fork or what.

3:51:59 beach I don't think I put any such directory in SICL.

3:52:08 beach And I haven't used your fork yet.

3:52:08 drmeister I thought the fork would be useful because I can push changes into it and then you can compare the changes to your code and decide whether they are a good idea or a bad idea.

3:52:41 beach At the moment, I don't know GIT well enough to do that.

3:53:07 drmeister Hmm, well then I'm confused. When I fetched your upstream changes and rebased them into mine there were messages about the "meister" directory. I must have misinterpreted them.

3:53:37 beach But I told you already that the PRINT-OBJECT method specialized to a condition is not how I would have done it.

3:53:51 drmeister Good, well, that's one less thing to have to wrap my mind around. It's so flexible at this point I can tie it in a bow.

3:54:19 drmeister No sweat - I'll take that out.

3:54:31 beach However, because you did create that method, I am now considering putting in enough infrastructure for internationalization.

3:54:39 drmeister What was the problem with what I did? You said there were constraints on print-object for conditions.

3:55:02 beach drmeister: Yes, hold on. I'll find the Common Lisp HyperSpec page for you...

3:55:14 drmeister internationalization? Did I use a Canadian spelling or something?

3:55:34 stassats considering that :report for conditions is specified to use print-object

3:55:50 beach stassats: Yes, but there is a condition in there.

3:56:20 beach clhs define-condition

3:56:21 specbot http://www.lispworks.com/reference/HyperSpec/Body/m_defi_5.htm

3:56:32 stassats a condition in condition printing?

3:56:45 drmeister Was "internationalization" a result of autocorrect? Or do you mean to support different languages for error messages?

3:57:11 stassats * wonders what word could be corrected to "internationalization"

3:57:14 beach drmeister: The latter.

3:57:48 beach drmeister: Condition reporting tests *print-escape*.

3:58:07 beach (if *print-escape* (call-next-method) ...)

4:00:03 beach drmeister: Do you see it? Under :REPORT near the middle of the page.

4:01:33 beach drmeister: Did you faint?

4:01:54 drmeister I'm trying to figure out what the point of not printing conditions when *print-escape* is.

4:01:59 drmeister nil.

4:02:06 drmeister Or do I read that incorrectly.

4:02:26 beach Presumably, you get the #<condition...> syntax then.

4:02:32 drmeister Other way around.

4:03:00 drmeister Oh, I think I see.

4:03:16 drmeister But is that a constraint on defining print-object for conditions?

4:03:27 beach I suppose not.

4:03:29 drmeister Other than don't expect *print-escape* to be anything other than nil.

4:04:07 beach But it would be recommend to define print-object methods as if the :report had been given.

4:04:07 drmeister Hang on, sometimes I have to read things several times to get it.

4:06:44 drmeister I see, I'm confusing the :report option with the print-object method. They can be two different things. Providing an argument to :report generates a print-object method and then explicitly providing a print-object will override that method.

4:07:20 beach "Replace" I would think.

4:07:41 beach ... rather than "override", I mean.

4:08:14 drmeister Or do I still have that wrong. You are saying that a :report can be given AND a print-object can be provided. Or does the print-object specialized to a condition replace what is generated by the :report option?

4:08:22 drmeister I can look this up myself - hang on.

4:08:37 beach The latter I think.

4:11:05 drmeister In ECL CLOS there is a print-object method specialized to CONDITION. That method looks up the 'report-function slot of the condition and if its a string it prints it and if its a function it funcalls it.

4:11:56 beach That doesn't sound conforming to me. Maybe others can help.

4:12:15 stassats no, that's exactly what it should do

4:12:26 beach :(

4:13:10 drmeister So a print-object method specialized to a condition derived from CL:CONDITION will be called first and if it invokes call-next-method then the default behavior specified by the :report option will be carried out.

4:13:46 stassats drmeister: nope

4:14:54 drmeister Nope? What did I get wrong?

4:15:27 stassats ok, everything, i was saying that it should do only about the last part of your sentence

4:15:29 beach In SBCL, when I do (define-condition bla () ()) and then (make-condition 'bla), the result is printed using the #<BLA ...> syntax. *PRINT-ESCAPE* is T

4:15:40 stassats i.e. calling a function or printing a string

4:15:53 stassats i didn't realize that by CONDITION you meant CL:CONDITION

4:16:20 drmeister stassats: Which sentence? I have a tendency to go on and on.

4:17:20 oleo ** NICK Guest76007

4:18:46 drmeister Ok, it's ok. I always have the source code.

4:18:55 beach drmeister: It looks to me like the default method on PRINT-OBJECT specialized to CONDITION should use PRINT-UNREADABLE-OBJECT.

4:20:15 beach stassats: Are you saying this is wrong?

4:21:13 drmeister If *print-escape* is T then it invokes (call-next-method) which probably invokes PRINT-UNREADABLE-OBJECT.

4:21:15 drmeister https://www.irccloud.com/pastebin/zpvSRZXV

4:22:22 stassats either way will be equivalent to what clhs describes, as long as you don't expect the method dispatch that clhs describes

4:23:17 stassats as long as you don't write any print-object methods yourself, you can't notice a difference

4:23:23 stassats but if you do, and call-next-method, you will

4:24:29 drmeister Got it.

4:24:31 drmeister Thank you.

4:25:33 beach drmeister: So in ECL/Clasp, when you define a new condition with a :REPORT init, no new method is created?

4:26:56 beach s/init/in it/

4:26:59 beach Damn!

4:26:59 drmeister I don't think a new method is created. The only place print-object is mentioned in the conditions.lsp source file is in that (defmethod print-object ((c condition) stream) ...)

4:27:11 beach drmeister: I see.

4:27:56 drmeister The :report argument gets put into the 'report-function slot and that print-object method specialized to CL:CONDITION reads that slot.

4:28:09 nyef Mmm. CONDITIONs aren't fully integrated to CLOS, for a variety of historical reasons.

4:31:23 loke Xach: assuming you're reading messages after you wake up: I've created the two bug reports on zs3 now.

4:31:28 loke https://github.com/xach/zs3/issues/14

4:31:34 loke https://github.com/xach/zs3/issues/13

4:31:52 beach If I do this: (define-condition bla () () (:report "hello"))

4:32:03 beach And then: (compute-applicable-methods #'print-object (list (make-condition 'bla) *standard-output*))

4:32:17 beach I would fully expect a method specialized to BLA.

4:32:58 beach SBCL does not do that, and apparently ECL/Clasp doesn't either.

4:34:57 beach "Specifying (:report report-name) in the definition of a condition type C is equivalent to: ..."

4:36:14 drmeister I guess "equivalent to" is not equivalent to "exactly the same as".

4:36:29 beach I think it is.

4:38:13 beach But at least I now understand the reason for the different interpretations of the Common Lisp HyperSpec.

4:44:32 nyef beach: CLHS 9.1.3 also seems relevant, plus a restriction "* The :report option of define-condition, not defmethod for print-object, must be used to define a condition reporter." from the page for Condition Type CONDITION.

4:46:38 beach nyef: Indeed.

4:47:30 beach So they are equivalent, but you aren't allowed to use the second one.

4:47:33 beach Right?

4:47:59 stassats the user

4:48:09 beach Yes.

4:49:26 beach This discussion deserves a chapter in my book "Common Lisp for Language Implementers".

4:50:53 beach drmeister: So I was wrong about the restrictions you have to respect when writing a method on PRINT-OBJECT for reporting those conditions. You aren't even allowed to write such a method.

4:54:19 nyef Hrm.

4:54:36 nyef beach: You might also be right in your expectation vis-a-vis compute-applicable-methods, though.

4:55:02 beach Yes, I still think that.

4:55:26 beach That issue seems orthogonal to this one.

4:55:33 nyef It'd be another matter if print-object were defined to be a generic function rather than a standard-generic-function.

4:55:54 beach Why?

4:56:22 beach Oh, because then you might have some other mechanism for creating methods?

4:56:42 nyef Pretty much, yeah.

4:56:48 beach Yeah, I think I see.

4:56:59 stassats nyef: aren't all gfs in the standard standard?

4:57:38 nyef stassats: I wouldn't know either way, but compute-applicable-methods only has behavior defined for standard generic functions.

4:57:58 stassats you can't create non standard without MOP, can you?

4:58:06 stassats mop c-a-m

4:58:07 specbot compute-applicable-methods: http://metamodular.com/CLOS-MOP/compute-applicable-methods.html

4:58:29 nyef The MOP isn't required to apply when we're discussing interpreting CLHS.

5:01:48 murftown hey, how can I see a line number for my errors in sbcl?

5:02:17 beach murftown: You use SLIME and then press "v" on your stack frame. You don't use line numbers.

5:03:03 murftown beach: ahh, does slimv do that too?

5:03:14 beach No idea. Sorry

5:03:25 murftown beach: '(no worries)

5:04:32 beach This is how I am thinking of doing internationalization: Define a LOCALE class containing lots of things like printing dates, but also an instance of the class LANGUAGE. There is a *CURRENT-LOCALE* variable. There is a condition INTERNATIONAL-CONDITION with a :report of REPORT-CONDITION which is a generic function that is called with the condition, the stream and the value of (LANGUAGE *current-locale*).

5:05:26 beach Client code can use INTERNATIONAL-CONDITION as a superclass and write methods on REPORT-CONDITION specialized to their favorite language.

5:05:33 beach Opinions?

5:07:38 nyef ... I can't think through the condition and :report logic and whatnot right now, but one thing that does come to mind is that "localized-condition" might be a better name than "international-condition".

5:07:57 beach Yes, good. Nothing set in stone yet.

5:08:16 stassats nyef: localized to the error site?

5:08:19 beach Client code can of course choose not to use INTERNATIONAL-CONDITION in their own condition types. And they can override the behavior by creating a subclass of a condition that *does* use INTERNATIONAL-CONDITION as a superclass, just by providing their own :REPORT.

5:08:41 nyef Reason being that even within a nation, you can have different localization requirements (consider Quebec!), but you can also have multiple nations sharing the same locale in some sense.

5:09:01 beach Don't I know it.

5:09:19 nyef Mmm. You'd probably know it far better than I.

5:09:30 beach I am constantly trying to get dates printed in ISO8601, messages in English, and wind speeds in m/s.

5:09:31 Zhivago international-condition sounds like a really bad name.

5:09:50 beach Yeah, I get it. Thanks.

5:10:47 Zhivago One interesting point that you might have missed regarding i18n is that input and output can often have different localization requirements.

5:10:49 beach Zhivago: so is LOCALIZED-CONDITION better?

5:10:58 Zhivago No. It sounds like an earthquake.

5:11:02 beach Heh!

5:11:03 Zhivago Why not locale-condition?

5:11:06 drmeister I'm back - I had some computer problems.

5:11:19 stassats but people in the same city can speak different languages, so, how localized is it?

5:11:21 beach Zhivago: Thanks for the hint about input and output.

5:11:22 Zhivago Since it seems to be about locales being defective of something.

5:11:25 nyef Zhivago: Because it sounds like a problem with the locale system?

5:11:39 Zhivago What's the condition about then? Perhaps I missed something.

5:12:13 Zhivago Yeah, you might want to e.g., input in English, but output in Chinese.

5:13:01 beach How about I just call it cleavir-i18n:condition?

5:13:03 Zhivago Or the reverse -- e.g., to search for Chinese text but display in English.

5:13:21 Zhivago That sounds reasonable.

5:13:42 beach Good. Thanks.

5:14:11 beach drmeister: You'll have to catch up by reading the logs. Very interesting discussion.

5:15:10 beach I feel my day has already been very productive, even though it has lasted for only 2 hours so far.

5:15:25 Zhivago Hmm, I guess anything you do using specials for i18n is doomed to probably suck pretty hard, regardless.

5:16:21 nyef Okay, it's late, I'm going to wish you all progress on your projects and luck wherever you need it, and head to bed.

5:16:31 drmeister I see print-object on conditions is verboten

5:16:40 beach Seems that way, yes.

5:21:56 drmeister beach: When I (asdf:load-system :cleavir-ast-to-hir) the :cleavir-env package doesn't get defined. Is that to be expected? Should I also need to (asdf:load-system :cleavir-generate-ast)?

5:22:11 drmeister I know I should be using uninterned symbols.

5:24:04 drmeister I'll load both systems.

5:25:10 beach It should be automatic. Probably a :depends-on missing.

5:25:39 beach No wait. An AST can come from elsewhere.

5:26:03 drmeister Got it.

5:26:34 drmeister * is pleased with himself that he got ASDF working.

5:26:50 beach * vanishes for about half an hour.

5:27:01 drmeister (asdf:load-system :cleavir-generate-ast) (asdf:load-system :cleavir-ast-to-hir) works now - thanks.

5:30:42 drmeister beach: When you get back I tried: (setq ast (cleavir-generate-ast:generate-ast '(let ((x 1)) (+ x x)) *clasp-env*))

5:30:53 drmeister (cleavir-ast-to-hir:compile-toplevel ast)

5:31:13 drmeister I got an error "illegal successors: NIL"

5:32:29 drmeister I'll be up for a little while longer if you get back in time. Otherwise I'll talk to you tomorrow.

5:51:45 beach drmeister: I'll look into it. Like I said, not tested yet.

5:51:56 beach That's today's task.

5:54:14 drmeister You were generating HIR a while ago. Is the expression above using the new multiple value return stuff you just added?

5:54:20 murftown hey so you know how #'foo represents the function foo not the variable foo? How can I get #'somevar where somevar gets evaluated to a symbol name?

5:54:46 beach drmeister: I completely changed the AST-to-HIR code.

5:54:56 drmeister Ohhhh. Ok then.

5:55:29 beach murftown: You are confused.

5:56:28 beach murftown: If you just type somevar, it means the variable with that name. It doesn't get EVALUATED TO the variable with that name. When evaluated, it returns the VALUE of the variable with that name.

5:57:17 murftown beach: but that's what I have is somevar whose value is a SYMBOL, i.e. a potential name for a function

5:57:41 murftown sorry if question was unclear

5:58:12 beach murftown: Let me get this straight. You have a variable, and the name of that variable is a symbol. That symbol is the name of a function. Given the variable, you want to return the function?

5:58:39 murftown I have a variable, and the VALUE of that variable is a symbol.

5:58:50 beach Sorry, yes.

5:58:52 beach murftown: (fdefinition somevar)

5:58:53 murftown That symbol is the name of the function, and now I want to call the function

5:59:12 beach (funcall (fdefinition somevar) arg1 arg2 ...)

5:59:48 murftown beach: perfect, thanks!

6:00:03 beach Anytime.

6:00:55 drmeister Well, bonne nuit!

6:01:30 beach Good night drmeister.

6:05:50 stassats murftown: (funcall somevar arg1 arg2 ...)

6:06:21 murftown stassats: thanks, beach gave me what I needed with fdefinition

6:06:30 murftown I can't believe how powerful lisp is!

6:07:57 beach Glad you like it. :)

6:09:29 stassats murftown: except that you don't need fdefinition

6:13:40 murftown stassats: well it solved my problem. I already tried (funcall somevar arg1 arg2) but it didn't use the function namespace for somevar

6:13:55 stassats that's not right

6:14:16 stassats it does exactly the same thing as (funcall (fdefinition somevar) arg1 arg2)

6:21:12 murftown stassats: ok, I see it doing mostly the same thing, but I still can't fix my code to work without fdefinition for some reason

6:23:49 loke murftown: (funcall #'somevar) ?

6:23:59 beach loke: No.

6:24:02 stassats that's not what fdefinition does

6:24:12 loke Ah right. Thats FUNCTION

6:24:12 beach loke: The value of the variable is the name of the function.

6:24:25 beach murftown: Don't worry about it. We'll teach you about "designators" in due time.

6:25:22 beach drmeister: That example now works for me.

6:25:38 loke let me guess then: somevar contains a lambda list?

6:26:15 stassats that's not what fdefinition does

6:26:33 murftown beach: sounds good. if interested I'd be happy to show you guys what my use-case is here. I created a macro called cpnd-funcall that allows me to input a list as a "function name", and it will call the function with the same name, i.e. (cpnd-funcall '(hello func with compound name)) will call function #'|(hello func with compound name)|

6:27:18 loke murftown: that sounds... weird

6:27:30 murftown loke: :) it is

6:50:23 beach Time to get to work!

6:57:20 murftown how can I turn a list (a b c d e) into one where each element is quoted, ('a 'b 'c 'd 'e) ?

6:59:23 loke murftown: are you sure you understand what you're asking?

6:59:35 murftown loke: yes

6:59:43 loke you understand what quoteing does, right?

7:00:06 Zhivago Perhaps he means '(a b c d e) -> ('a 'b 'c 'd 'e).

7:00:07 loke It prevents evaluation. If you have a list (a b c d e) there is no evaluation going on, unless you pass it to EVAL

7:00:24 murftown loke: I get that

7:00:31 loke Zhivago: Well, yes. That's easy

7:00:51 murftown here's what I came up with: (defun quoteall (l)

7:00:51 murftown (mapcar (lambda (i) (list 'quote i)) l)

7:00:52 murftown )

7:00:57 loke if you _really_ wan tit, just try (mapcar #'(lambda (v) (list 'quote v)) the-list)

7:01:09 murftown loke: exactly

7:07:15 Zhivago There's also `',i :)

7:08:28 murftown Zhivago: whoa, really! I was looking for something like that but I can't wrap my mind around that yet

7:15:39 Zhivago 'x is (quote x), so ',x is (quote ,x), so `',x is `(quote ,x)

7:20:13 loke I have a need to do basic source code highlighting in Coomon Lisp. Anyone know of any packages that can do this?

7:20:27 loke There is one that can do Common Lisp, but I'd like to do multiple languages

7:20:36 joga emacs? :)

7:20:46 loke This is for a web application

7:20:56 loke It's already written in CL :-)

7:21:05 H4ns doesn't lisppaste contain something like that?

7:21:32 loke H4ns: it does. you're right

7:22:05 loke I want to integrate it into my simple markup processor.

7:22:05 loke https://github.com/lokedhs/cl-markup

7:25:17 loke Where is the official source repository for lisppaste by the way

7:25:17 loke ?

7:28:54 |3b| lisppaste uses colorize

7:29:18 loke thanks

7:29:24 loke I thought that one only did CL

7:36:44 siwica I am currently learning lisp. What would be a good programming project that I could do to learn it?

7:37:32 Zhivago You could write an annoying irc bot.

9:18:52 wasamasa siwica: the best ideas usually come from scratching an itch

9:19:28 wasamasa siwica: like, if you notice some day that there's not a single decent epub reader for linux desktops, you go ahead and write one

9:20:04 siwica wasamasa: I have some experience with languages in other paradigms, so I wonderes if there are a few "canonical" beginners problems for functional languages.

9:20:11 wasamasa siwica: because you've been bothered enough by it to turn it into a project for yourself

9:20:24 wasamasa siwica: implementing map with reduce :P

9:20:39 wasamasa siwica: then, CL isn't particularly functional (out of the box)

9:20:45 wasamasa siwica: so that would make little sense

9:21:12 siwica wasamasa: what do you mean its not functional

9:21:16 siwica I am using emacs lisp in particular

9:21:27 wasamasa emacs lisp is not functional programming either

9:21:56 wasamasa in fact, I'd call it the lisp dialect furthest away from FP, simply because you're forced to modify global states to get anything meaningful done :P

9:22:26 wasamasa siwica: it has little support for functional programming

9:23:23 wasamasa siwica: if you want to discuss other lisp dialects than common lisp, I encourage you to strike up a conversation on ##lisp or the respective channel of that lisp dialect (for elisp, that would be #emacs)

9:23:24 |3b| siwica: this channel focuses on common lisp (and interprets 'lisp' on its own as common lisp), which supports lots of programming paradigms to varying extents

9:24:17 siwica ok, sorry

9:24:36 wasamasa siwica: if you want to learn a language that's supporting FP only, take a look at haskell

9:25:04 siwica Yes I did

9:25:21 siwica that would be my prefered language

9:25:35 wasamasa /join #haskell then

9:25:36 siwica but i thought since I am using emacs anyway, I might aswell...

9:27:08 wasamasa you could go and write some more coherent haskell support for emacs, yes

9:27:19 wasamasa but that has pretty much nothing to do with functional programming

9:29:40 |3b| from what i've seen, people talking about function programming with "lisp" usually mean some subset of scheme

9:29:45 |3b| *functional

10:12:59 didi Is there a convention for using comments on top of lisp libraries, specially single file ones?

10:29:59 chaotik ** NICK shortCircuit__

10:45:14 Shinmera Ah, I missed you, old friend. http://shinmera.tymoon.eu/public/screenshot-2014.11.19-11:44:48.png

11:16:18 pjb minion: memo for Shinmera: since setting an undeclared variable is implementation dependant, an implementation could define automatically _lexical_ variables for them, so you may have fun at the REPL and keep your programs working. But this would need additionnal code in the implementation.

11:16:18 minion Remembered. I'll tell Shinmera when he/she/it next speaks.

11:43:23 pjb Zhivago: beach set up a lisp project list! http://metamodular.com/Common-Lisp/suggested-projects.html

12:09:11 schjetne pjb`: thanks for the link that's relevant to my interests as well

12:12:44 alpha- pjb` interesting site

12:13:30 Shinmera pjb`: I'm still confused a bit. If I say define a function that uses some variable that is neither lexically bound nor declared special and then later use defvar, are the consequences implementation dependant?

12:13:30 minion Shinmera, memo from pjb: since setting an undeclared variable is implementation dependant, an implementation could define automatically _lexical_ variables for them, so you may have fun at the REPL and keep your programs working. But this would need additionnal code in the implementation.

12:18:31 francogrex hi, in a program some arguments are passed as keyword: for example (r% "rownames" NONRBIND :do.NULL nil) but the case of the keyword matters for example: ;R! Error in function (x, do.NULL = TRUE, prefix = "row") ;R! unused argument(s) (do.null = FALSE)

12:19:32 francogrex as you can see I must pass do.NULL as it is and not do.null as it is being translated to be passed to the receiving app. Can it be done to "fix" the case of the keywords

12:20:30 |3b| pjb: is accessing non-existent variables specified somewhere to be implementation dependent as opposed to undefined?

12:20:49 |3b| * thought it was just completely ignored by the spec, so presumably undefined

12:22:05 |3b| francogrex: did you want to use :|do.NULL| ?

12:22:54 |3b| or look up the correct case in a table somewhere if you know them all in advance

12:23:44 |3b| other option is tricks with reader case, but unless you have a modern-mode lisp that would probably be annoying too

12:24:14 |3b| (either you would have to type everything else in upper case, or you would have to type :DO.null to get "do.NULL")

12:26:32 francogrex |3b|: :|do.NULL| does not do the trick unfortunately

12:26:51 |3b| well, you also need to print it correctly, which "do.null" suggests you aren't

12:27:08 |3b| with default settings, :do.NULL is read as a symbol named "DO.NULL"

12:27:38 |3b| so if it prints as "do.null" you are discarding any case info in the symbol, and putting 'correct' case in the symbol name won't help

12:29:35 jusss` ** NICK jusss

12:37:05 alpha- http://metamodular.com/Common-Lisp/lispos.html

12:37:14 alpha- this looks interesting

12:37:17 alpha- and crazy

12:49:11 francogrex my issue here is (equal :a :A) => T I want it to give nil

12:49:23 joga regarding crazy + os... this is pretty neat http://www.templeos.org/

12:49:36 Xach francogrex: why?

12:51:37 francogrex Xach: I am passing args keywords and the app wants them in a specific case: (r% "rownames" NONRBIND :do.NULL nil) => ;R! Error in function (x, do.NULL = TRUE, prefix = "row"): unused argument(s) (do.null = FALSE)

12:52:24 Shinmera What kind of application cares about the keyword case

12:52:28 francogrex R

12:52:29 alpha- they propose tag filesystem instead of hierarchical - you can do thi in linux if you wanted

12:52:40 francogrex a lot of others as well

12:53:03 Shinmera so you're interacting with another programming language, not a lisp application

12:53:31 alpha- they propose unified memory with ram ascache but the point us unclear since linux kernel already caching read data in ram

12:53:48 francogrex yes another

12:54:19 Shinmera In that case it's your wrapper's job to translate the keywords into the proper case.

12:54:27 alpha- they complain that serialization/deserialization is a prolem - is it really

12:55:25 H4ns alpha-: if it does not appeal to you, you can safely ignore the proposal

12:55:25 Shinmera alpha-: If the three gazillion different file formats are not a good enough proof that serialization is a problem, I don't know what is.

12:57:14 |3b| francogrex: either do (eql :|a| :A) => nil, or configure the reader to preserve or invert case

12:57:21 alpha- H4ns it is appealing

12:57:26 alpha- idea

12:57:35 alpha- but not for any of the reasons listed on that page lol

12:58:04 |3b| francogrex: just keep in mind that preserving case means you need to type all CL symbols in upper case, and inverting case will invert case which you would need to undo when printing

13:00:58 Shinmera Or you have to wrap it. (defun r-rownames (&key do.null) (r% "rownames" NONRBIND :|do.NULL| arg))

13:01:02 francogrex |3b|: ok

13:01:13 Shinmera *arg=>do.null

13:01:17 alejandrozf hi everybody, I have a question, What you think is better CL or Racket?

13:01:26 Shinmera This channel is about CL, so

13:01:39 Shinmera It's either off-topic or a trivial answer.

13:01:51 alejandrozf ohh sorry

13:02:03 Shinmera You can try ##lisp for general lisp discussion

13:02:10 |3b| there is also #racket if you'd prefer the other trivial answer :)

13:02:16 |3b| (or #scheme)

13:02:35 alejandrozf i'm so newbie, I'been exploring both but I didn't decided

13:02:55 Shinmera I'm sure there's nothing wrong about learning both.

13:03:20 |3b| most of the arguments are personal preference, so use whichever you like best

13:03:48 |3b| at least for CL vs scheme

13:03:57 alejandrozf I started with CL but GUI its difficult to me...

13:04:06 wasamasa webui ftw

13:04:16 |3b| 'racket' implies a specific scheme implementation + extensions and extra libraries, so not quite as direct a comparison to CL as a language

13:04:17 wasamasa alejandrozf: also, discuss such stuff on ##lisp

13:04:29 wasamasa alejandrozf: #lisp is about common lisp only

13:04:38 alejandrozf ok!

13:21:25 francogrex |3b|: you expected :do.Null to be transmitted as :DO.NULL but I think the programmer has placed a code that converts the cases to downcase before passing it on to R: this in his source: (string-downcase (substitute #\_ #\- (symbol-name arg)))

13:22:26 Shinmera francogrex: in that case no matter what kind o symbol is read it's not going to end up being the right case.

13:22:35 |3b| right, that's what i was saying, which is why :|do.NULL| doesn't help

13:24:15 francogrex true

13:24:35 Shinmera Looks like you need to fix the library.

13:24:50 francogrex the programmer assumed all args are downcase because in general they are, except for that 0;1% situation

13:26:26 rvirding__ ** NICK rvirding

13:34:50 nortti ** NICK lawspeaker

13:36:22 lawspeaker ** NICK nortti

13:56:44 VargaD_ ** NICK VargaD

14:22:09 nyef G'morning all.

14:22:11 schjetne wasamasa: no love for the Zetalisp or Maclisp communities? ;)

14:23:25 schjetne I've been thinking of doing a toy Lisp 1.5 implementation as a bit of historical re-enactment

14:30:43 wasamasa schjetne: don't know enough about those to be honest

14:33:49 schjetne The predecessors to Common Lisp, but I guess any legacy code still in use has been ported to CL

14:34:45 schjetne That was one of the objectives of CL, to make transition from older dialects as smooth as possible

14:35:37 schjetne If you're interested, check out Gabriel & Steele - The Evolution of Lisp

14:38:56 bambams__ ** NICK bambams

14:45:42 pjb` ** NICK pjb

14:48:15 pjb |3b|: it's undefined. Implementations are free to implement whatever they want in this case: signal an error, create a special variable, create a lexical variable, launch a ballistic missiles onto the programmer, whatever.

14:48:59 pjb |3b|: most implementations seem to find useful to create a special variable. I would argue that it would be more newbie friendly, and otherwise REPL user friendly, to define a (global) lexical variable.

14:49:43 pjb Shinmera: yes, if you define a function with a undefined free variable, and later define it, it is undefined, implementation dependant, what may happen.

14:49:58 pjb There's no forward definition for variables.

14:52:12 drmeister Morning nyef

14:52:13 pjb alpha-: this is not htat crazy: http://wiki.osdev.org/Main_Page http://micropenguin.net/files/Other/OS/os-dev.pdf

14:55:08 pjb minion memo for francogrex: cf. also my emacs commands upcase-lisp and upcase-lisp-region.

14:56:02 Shinmera pjb: Alright, thanks!

14:57:07 pjb schjetne: no love here, but try ##lisp.

14:59:57 pjb schjetne: legacy code doesn't have to be converted to run in Common Lisp in a lot of cases. cf. eg. http://www.informatimago.com/develop/lisp/com/informatimago/small-cl-pgms/wang.html.in

15:00:37 pjb schjetne: but of course, if you want to make it evolve, you would better upgrade it to CL.

15:00:45 axion how do i push an element onto an adjustable #2a?

15:00:52 pjb schjetne: old code usually use old style that is rather unmaintainable.

15:02:10 pjb axion: you don't.

15:02:38 pjb You store the element with (setf aref), or you adjust-array the array. You can't do both at the same time.

15:03:12 pjb Of course you can wrap both operations in a single function of yours, but notice how adjusting a 2D array in general will not create a single new slot, but a whole column or a whole row!

15:03:59 axion aha ok.

15:05:20 pjb schjetne: Notice however, that sometimes some porting may be needed. One thing is that in old lisps, bindings are always dynamic, so you often have to add special declaration (or thru some macrology, since they're expected to be implicit). Another problem is when you're dealing with an old lisp that is a lisp-1 (again solvable thru macrology).

15:06:18 pjb schjetne: and finally, in some old lisps, some datatypes allow accessors reserved nowadays for different types. Notably, (CAR symbol) and (CDR symbol) (with the corresponding RPLACA/RPLACD) were meaningful in some lisps! (cdr symbol) would give the plist.

15:07:27 pjb But I would say that such macrology is good as long as you only want to run a small legacy function. If you had a big legacy system using a lot of those dialect specific features, you'd want to go metalinguistic, and re-implement the old lisp (or indeed, port the code).

15:07:59 pjb Notice how Norvig re-implemented the algorithms in Common Lisp in PAIP, instead of running directly the old code.

15:08:13 schjetne Ah, that's another book on my reading list

15:08:48 schjetne When you say the cdr of a symbol gives a plist, is that a property list of things bound to the symbol?

15:09:01 pjb Yes.

15:09:04 pjb clhs symbol-plist

15:09:05 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_symb_4.htm

15:09:23 pjb But in old lisps, the symbol name, the symbol value, the symbol functions, etc, were stored on the plist too.

15:10:30 drmeister Does anyone use "magit" on emacs? It opens COMMIT_EDITMSG to edit a message - how do I save/close that buffer to continue the commit? I'm committing Common Lisp code.

15:10:45 pjb I use normal vc.

15:11:06 pjb Mostly: C-x v v

15:11:52 oleo drmeister: normally via C-x C-s

15:11:56 oleo afaik....

15:12:00 schjetne pjb: I haven't really looked much into all the different symbol cells, I should probably do that.

15:12:08 oleo just like normal....buffer/file

15:12:48 pjb Normally, yes you need to save and close the buffer to perform the commit.

15:18:51 axion when vector-push-extend needs to resize the underlying storage, how does it do so? increase by 1, by initial storage length, or?

15:20:30 Shinmera !clhs vector-push-extend

15:20:33 Shinmera clhs vector-push-extend

15:20:33 specbot http://www.lispworks.com/reference/HyperSpec/Body/f_vec_ps.htm

15:20:52 Shinmera "vector-push-extend new-element vector &optional extension" "Extension is the minimum number of elements to be added to vector if it must be extended."

15:23:55 axion ah, implementation-dependent