| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Comments:
<0> rog_: Right. The CMUCL internals manual has a very detailed section on optimization of "local calls."
<0> nyef: Because most programs don't redefine things after they've started runnning. I'm sure you could create a system that allowed redefinition, but it'd be much more complex.
<1> adeht: interface checking holds the possibility of doing the same kinds of optimisations across modules that are now done within a module. to a degree.
<2> quotemstr:you can simply add a check for redefinition before the thing you'd normally do w/o redefinition.
<3> nyef: perhaps you could add asimple test to inhibit the "redefinition" warning when it comes from the same file.
<4> pjb: Hasn't most of the groundwork for that in SBCL been laid by now?
<0> pkhuong: And take a hit on every invokation of that thing, all to support functionality that isn't used in a deployed program.
<2> quotemstr: there is next to no cost on modern arhcitectures.
<0> nyef: Yes, much to my annoyance in SLIME. :-)
<1> apart from anything (and this is where the thread started) an explicit interface makes a very useful adjunct to the documentation.
<0> pkhuong: If you have the compiler inline a function, you have to keep track of where it's been inlined so those functions can be recompiled after the original function has been changed.
<2> quotemstr: the 90s want their non superscalar architectures back.
<5> hm how do i undefine method that i created with defmethod?
<0> pkhuong: CMUCL keeps track of that automatically, according to the manual, but it's still extra complexity.
<4> stefan-: Pop the slime inspector on the generic function.
<6> stefan-: the easiest way when using slime is to inspect the generic function
<6> C-c I #'my-generic-fun RET
<7> rog: How is that? Proper documentation already mentions the parameters.
<5> cool! i had not seen the inspector!
<0> Woah, that is cool.
<1> adeht: but proper documentation doesn't always exist. which was where this thread came from...
<4> rog_: That's not really a deficiency which should be corrected in the language.
<0> Speaking of slime -- does a GDB-style debugger for lisp exist?
<4> quotemstr: What, one that ****s horribly?
<7> rog: Then the usefulness of the library is doubtable.
<1> adeht: and i've seen a great deal of lisp documentation that doesn't mention the return type or default parameter values.
<5> i'm getting Attempt to add the method #<STANDARD-METHOD MARKUP (T T) @ #x10857e7a> to the generic function #<STANDARD-GENERIC-FUNCTION MARKUP> but the method has more required arguments than the generic function.
<7> rog: Well, "default parameter values" are there for when you don't _care_ what they should be.
<6> stefan-: if you would like to redefine the gf entirely, use (makunbound 'markup)
<6> oops, (fmakunbound 'markup)
<4> adeht: Not necesarily.
<1> adeht: perhaps. but people use (for instance) sb-introspect but it has no documentation (AFAIK)
<7> nyef: I know. There is another use, but I picked one for consistency.
<4> adeht: You usually care about :direction :input when you use with-open-file, but since it's the default you don't have to specify it.
<5> #<STANDARD-GENERIC-FUNCTION MARKUP> is not a valid function spec
<5> the problem is that i defined that method as (defmethod markup ((c component)) ...)
<5> which is wrong of course, but now i can't seem to get rid of it :)
<7> nyef: I always specify :direction :input if I need it.
<6> stefan-: (fmakunbound 'markup)
<5> Xach, i tried, i get that last error i posted
<6> stefan-: Hmm, are you sure you typed what I typed?
<4> stefan-: No you didn't. fmakunbound doesn't do that.
<5> CL-USER> (fmakunbound #'markup)
<5> oh ****
<7> nyef: And this situation is a bit more "blurry" here because the CL standard is quite stable, in opposition to a non-standardized library.
<6> stefan-: you son-of-a-goat!
<5> grrrr!
<5> beeehhh
<4> adeht: And the CL standard is fairly well documented, which is rather the point.
<5> is this a common pattern: (defmethod markup ((c component)) (error "Must override markup"))
<4> stefan-: That, or just letting it throw a "not-implemented" error.
<4> Err... No applicable method, rather.
<8> You guys wanna see what I've coded since yesterday? It's a blog system, my first "real" Lisp application that does something.
<5> nyef, by not defining markup at all?
<9> write a function that replaces "some" atoms of a list with other atoms.Does this mean using rplaca or returning a new list?
<6> hedos: yes!
<6> stefan-: define it as a generic function without methods.
<4> stefan-: Yeah, just use defgeneric for it.
<8> Great! The page runs on my laptop / modem cable: http://70.81.253.46:1450/blog
<5> ah right. then i can still add some documentation to it.
<6> hedos: i can't access odd ports :~(
<4> Using defgeneric before defmethod has some semantic differences anyway, and keeps SBCL from complaining.
<10> Xach: but it's even!
<8> Aye, my ISP won't let me serve web page on the standard port :(
<11> stefan-: in fact it's the way to specify documentation to a generic function. the docstring you p*** to defmethod is local to the method and not to the function and is not as easily retrieved.
<4> hedos: Can we interest you in some virtual hosting with a proxy that will do the port mapping for you?
<2> hedos: 8080?
<8> 8080 isn't allowed either.
<6> michaelw: not even enough
<8> But a proxy might very well interest me.
<12> better.
<7> hedos: Looks pretty cool :)
<13> worse.
<4> Worse is better?
<10> Xach: that's odd...
<8> nyef: Do you know where could I find such available proxies?
<0> hedos: Get some shared hosting. It can be very cheap.
<4> hedos: I believe tech.coop provides such a thing... And they'll also let you run the Lisp code on their machines!
<7> hedos: I just hate the artifical width narrowing.
<8> adeht: thanks! the comments system is still very crude and unfortunatly there's no web interface to add comments yet.
<8> adeht: Ok, that's a good comment, I should have a look at what full width feels like.
<14> amazon's EC2 seems like an interesting hosting possibility.
<6> rainer joswig won't use it if you do.
<14> sometimes a lemonodor's got to do what a lemonodor
<14> er, lemonodor's got to do.
<14> we all have to make our own decisions. rainer has seemed to be a pretty reasonable person, i'm sure he has good reasons for his actions.
<15> Ah, a person-mutual-exclusive library
<0> It'd be nice with paredit-mode to somehow highlight the last ***p before point after paredit-close-parenthesis-and-newline so that you can tell when to stop hitting ')'
<16> m-x paren-mode?
<16> m-x show-paren-mode i mean
<17> quotemstr, can you elaborate? It does blink the matching parenthesis.
<0> _3b: Yeah, that's fine, but it doesn't highlight after )
<10> jlouis: hmm?
<17> (Either that or you don't have the most recent version of paredit.)
<0> Okay. Say we have (let ((function-name (or ***oc function-name +rename-funcs+) function-name|))), with | denoting the point.
<1> so many map functions, but... what's the most elegant way of doing the equivalent of mapcar on the values of a plist?
<0> Say I want to add another let binding after the first one --- I hit )
<0> rog_: I was actually wondering that myself the other day; I looked for a while in the CHLS, then switched to an list. :-P
<6> rog_: (loop for value on plist by #'cddr do (something-with value))
<0> Xach: That works too.
<6> oops, i err.
<1> Xach: not quite
<6> but that's the general idea.
<1> you mean (cdr plist), i presume
<6> (value)
<3> (loop for (key value . rest) on plist by #'cddr do (something-with value))
<16> quotemstr: hmm, guess not...make sure you have paredit 20 then? :)
<0> _3b: Does it highlight the ***p when you do that?
<1> i'd forgotten about on, by. i had had (loop for x = (cdr pl) then (cddr x) while x ...)
<18> what actions does R.J. have reasons for?
<19> hello, did anyone experienced that sbcl goes 100% after slime-disconnect?
<17> quotemstr, I fixed a bug in paredit 20 that kept it from blinking the matching parenthesis in certain obscure circumstances, or perhaps in a certain version of Emacs.
<0> rog_: Indeed. Xach's solution is elegant.
<0> Riastradh: Patch?
<6> elegance is in the eye of the brucio
<17> Patch? No, it's just up at <http://mumble.net/~campbell/emacs/paredit-20.el>;.
<17> (That file periodically changes -- indeed, it changed twice earlier today -- because it is not the finally released paredit 20, but rather still a beta.)
<18> is there a secret lispm v. lemonodor contest I failed to notice?
<0> Riastradh: Ah, I see. So should I update it daily?
<17> No, update it when I say so in #paredit.
<0> Riastradh: Why are paredit commands bound to M-up, M-down, but C-left and C-right?
<0> Ah, I see. :-)
<17> They do very different things. (-> #paredit)
<14> lichtblau: rainer doesn't like my postings about military robots.
<20> Not liking military robots sounds understandable.
<18> I see.
<18> (but that's what we're all reading planet lisp for!)
<6> there was also an ultimatum to me
<15> the lisp robots made an ultimatum to you or threatened to use military force?
<8> Is there a quick way to save to disk the content of a variable (which is a list of objects)?
<21> I think that depends on what information you really want to preserve.
<22> yes, if if you are willing to save the rest of the lisp state too
<0> hedos: Is WRITE sufficient for your purposes?
<20> hedos: you can use WRITE, or perhaps one of the serialization libraries
<8> Saving the whole state would be just good.
<23> hedos: i believe you're looking for cl-store (lisp's equivalent to python's pickel (more or less))
<4> Heh. "Stack-overflow detection on the value stack is transparent, automatic, and incredibly hairy." -- The OpenMCL documentation.
<8> cmm-, I'd be interested by your solution. How is it done?
<22> hedos: implementation-dependent. in sbcl, it's SAVE-LISP-AND-DIE
<8> Ok. Is there such a facility for CMUCL?
<0> hedos: Yes.
<24> hedos: (describe 'ext:save-lisp)
<0> hedos: It can't make an executable though.
<24> hedos: looking for cl-launch ?
<8> oh, great, thanks
Return to
#lisp
or
Go to some related
logs:
#linux
#suse
#sql
Buffalo terastation and smb4k
#linux
#perl
#bash
Reason -2 firmware udev debian
#lisp
#gentoo
|
|