| |
| |
| |
|
Page: 1 2 3 4 5 6 7
Comments:
<0> arr, the cvs man page ****s
<0> doesn't modules refer to the files you want to affect?
<1> fwiw, there are info docs on cvs
<0> the man page would be ok except it leaves out half the commands
<1> afaik, cvs modules can be files or directories; in a cvs server, I think it's possible that a module would be mapped to something in the server's filesystem not having the same name as the module's name
<0> tag doesn't seem to be in the info page either
<2> now i have 50 notes but i see what the problem is hahaha
<2> in effect, there is code that looks like: (let ((v nil)) (if v this that))
<2> harmless but the macro could be smarter about not generating unnecessary code
<3> rr--: how would the macro know that?
<3> sbcl is smart enough to figure out that (let ((x (+ 1 1))) (if (zerop x) a b)) has dead code, asking macro writers to do that themselves seems like a major issue.
<4> that's not actually what the macro does. it's rather (let ((v nil)) (if v body body)))
<4> so what you might want to do is (let ((v nil)) (flet ((foo () body)) (if v (foo) (foo))))
<5> moin
<6> segv: somebody with a code walker at hand could write a partial evaluator. That'd get rid of such cases :)
<3> michaelw: you know if were to start down that route i'd end up with sacla? :)
<3> s/if were/if i were/
<6> segv: life is tough, and then you save-lisp-and-die. ahem :)
<7> Why is the swank-scheme48 directory empty in SLIME CVS?
<3> z0d: because the scheme48 backend was removed from swank so that it could make modifications to swank itself which weren't neccessary for the common lisp backends.
<3> swank-scheme48 is a small form of SLIME.
<3> s/form/fork/
<2> segv: in case of DO-SEQUENCE macro, the expander can know at expansion time whether a particular keyword arg was or was not supplied. if not supplied, it can generate less code.
<3> rr--: you're quite right. i had missed the paste.
<7> segv: I see. The directory isn't removed though
<6> rr--: why don't you dispatch at run time on the type of sequence given? (etypecase seq (cons (dolist ...)) (vector (loop ... across ...)) ...)?
<6> rr--: also, then you could consider giving your do-sequence a type argument, a la CONCATENATE, and get rid of the run-time dispatch if it is given
<2> on q.1, it does dispatch at runtime but using COND not ETYPECASE ... on q.2 good suggestion
<8> hi moore33. How's work?
<9> Xof: Not bad, though it's ****ing hot in the office :)
<2> i guess ETYPECASE would be clearer
<10> sqweek, I remember only doing cvs tag whatevertagname .
<10> Or so
<0> that's about what i remember aswell
<10> Then I moved to darcs and all my problems disappeared in a puff of logic.
<1> trying to find out something about nconc - like should i : (let (a (b '(1 2))) (dolist (n b) (nconc a (list (1+ b))))) #| or |# (let (a (b '(1 2))) (dolist (n b) (setq a (nconc a (list (1+ b))))))
<1> not sure if the setq is necessary - not sure of (nconc a ...) would work to modify A when A was NIL
<3> gimbal`: the setq is neccessary.
<1> segv: ok; danke
<3> destructive does _not_ mean that the parameter si replaced with the result, only that the parameter may be modified.
<3> this goes for all the other destructive function like delete as well.
<1> mumble mumble you can't change the CDR of NIL heh
<1> and like the CDR of NIL I guess is also the LAST of NIL - realizing, that represents some frobbled lisp grammar
<1> like, I always wondered why those destructive fns wouldn't work when the first arg was NIL; I think that's it
<1> ^ that's what the mumble mumble wasn't
<1> well, would work, but wouldn't work to 'modify' the NIL arg value
<1> line-by-line communiction doesn't work to make paragraphs well
<8> work on getting the lines comprehensible, and then the paragraphs will take care of themselves
<1> makes sense to me bob
<1> (let (a) (boundp 'a)) ; => NIL ? wouldn't A be bound in the let spec ?
<11> The function bound determines only whether a symbol has a value in the global
<11> environment; any lexical bindings are ignored.
<1> oh; thank you
<11> s/bound/boundp/
<8> how's xcl, piso?
<3> gimbal`: which isn't to say you can't do that, just that it's going to me very non portable.
<11> Xof: it p***es 10177 out of 10442 tests
<8> that's not bad
<11> proof-of-concept compiler, but no CLOS yet
<8> are you going to start with closette or pcl this time? (or lisp500? :-)
<11> seems to be reasonably fast
<11> definitely not pcl
<12> xclent
<11> I'm still looking at lisp500, but I'm not convinced it's really very correct (although the code is quite nice)
<13> when a depends on b and c, how do you call a and b and c ? (this is for the parameters of the procedure setting the dependencies) depender and dependee ?
<11> so, at the moment, I'm pretty much starting from scratch, with help from all of the available implementations ;)
<8> after a long time of being anti-PCL I think I'm beginning to come round to it again
<8> I wish I had time to actually measure the effect of a bunch of its optimizations
<11> my current plan, for what it's worth, is to start over with a different set of optimizations
<11> the main criterion being comprehensibility
<11> trying to understand pcl's optimizations makes my head hurt
<8> I think I've got them all into my brain now except for the permutation-vector stuff
<8> (I basically understand what a PV is and what it's doing, but I have no idea about the invariants that need to hold in various stages of internal computation...)
<11> I think I still have your documents in my gym bag :)
<11> but right now I'm just beginning the clos implementation, and I'm really only concerned with correctness and clarity
<11> (even though I'm normally quite fond of premature optimizations)
<14> How come there are not any common lisp implementations written entirely in common lisp?
<11> movitz is written entirely in CL
<8> sacla, too
<11> most other CL implementations need to interact with an OS written in C, so it's handy to write a bit of C code to glue things together
<1> would anyone happen to know of a cl-indent form for defgeneric, which would handle method-combination qualifiers correctly? (If I write any :around or otherwise-qualified methods in a defgeneric form, with the defgeneric indenter being used here, then emacs muxes up the indentation)
<11> sacla is incomplete to the point of not really running, however
<15> Hello.
<15> when a depends on b and c, how do you call a and b and c ? (this is for the parameters of the procedure setting the dependencies) depender and dependee ?
<8> "more depended-upon than depending"
<16> I had the same problem the other day, thanks Xof.
<2> sounds like A is a dependent of B
<2> so A is exempt from paying taxes, as long as B claims A as dependent i think
<2> which would make B the 'taxpayer'
<12> ingvar: I'm all for tanks running away from the player!
<17> rudi: It's... suboptimal. That means they're out of range of the player's missiles.
<18> Is there any way to tell when doing compile-file in slime which part of the code is giving me a package lock error?
<17> Though I did, for a while, have a hilarious bug. When teh enemy tanks fired their missiles, they were actually fired from the player's position. That was... amusing.
<19> ingvar: heh
<19> if you excuse the pun, I hope your game doesn't tank
<17> Well, in fixing that, I seem to have buggered missile-firing up, but I only had about 30 minutes of coding time this morning. We'll see where we are by tomorrow. The weekend will (alas) be dedicated to re-tiling walls.
<18> Ah, nevermind.
<12> ingvar: just rename the game "capture-the-tank" and be done with it :)
<12> (seriously, I'm looking forward to seeing it in action)
<6> hmm, is B~rge a female name? (.n
<20> What is the point of using the ";; -*- Package: FOO; -*-" construct in the first line of Lisp files, if the editor is smart enough to figure out which package the file is in based on (in-package) forms anyway?
<6> Athas: the editor could be less smart and it would work anyway? :) (i believe this is just historical baggage, FWIW)
<20> If an editor is smart enough to interpret the "Package:" stuff, I'd expect it to be smart enough to look for (in-package) forms in the buffer. :)
<20> Besides, it looks like this was done even on Genera, and I ***ume ZMacs was smart enough to scan the buffer for (in-package) forms.
<6> yeah, but it must parse more, understand comments, reader conditionals, etc.
<6> also, you could have your own macro which expands to an in-package form I guess
<20> That is the only valid case I can think of.
<10> But why have such things when you already know which package you should be in?
<10> (And you do know, otherwise you cannot write it in the Package: line)
<21> for evaluation/compilation commands to do the right thing, if i understand the question right.
<20> They won't necessarily do the right thing. The package specified in the Package line need not be the same package as in the (in-package) form, thus making the editor and compiler see different things.
<20> It seems like pointless redundancy to me.
<8> I'm not so certain that zmacs would scan the buffer for in-package forms
<10> Athas, it all depends whether "if the editor is smart enough" is true.
<21> well, if the editor understands the in-package forms that is
<10> +on
<10> Non-slime-using emacs, for one?
<20> If your Emacs is not using SLIME (or something similar), why would you care about the package? You wouldn't be able to symbolcomplete or eval forms anyway.
<21> i personally don't specify packages in the modeline; but some people still use LispMachines, you know.
<20> He can add the package specification himself if he wants to edit my code. :)
<21> actually i write modelines specifying the lisp mode for .asdf files, anyway.
<21> only.
<20> I can see the point of using Syntax or Mode options, but a respectable editor should be able to figure Package out on its own.
<21> i haven't written any real lisp for a too long time, too.
<10> Athas, to get a nice modeline? *shrug*
<20> Yeah, a long line of options does look very serious and respectable, even if all it does are senseless things like Base: 10. :)
<10> I mean the emacs modeline on the bottom
<19> also, a sufficiently smart compiler could style-warn that the in-package form and the modeline package don't match (:
<6> "sufficiently smart compiler" is something like Godwin's Law for Lisp, isn't it? :)
<22> what does save-lisp exactly do?
<20> The CMUCL function?
<7> "Sufficiently smart compiler" is an euphemism for "in theory".
<22> Athas, yes
<23> z0d: Which is, in turn, a euphemism for "not really"?
<20> Dumps a new corefile and dies.
<6> z0d: that's no contradiction
<20> So you can save a Lisp core with a given set of code loaded, and resume that core at a later time.
<7> nyef: Of course not. But the first form implies (a little) that such compiler exists.
<24> Hi. In UCW, would it be possible to create something like server-side image maps using the same kind of model that <ucw:a and <ucw:form use?
<22> Athas, what can i do with that corefile?how to load it back in?
<7> mc__: You can use it as a core file
<20> mc__: look at the CMUCL documentation (the manpage should be enough). It takes a parameter specifying which corefile to load.
<7> mc__: to start your Lisp
<19> I always use future-cl-v2::compile-file/sufficiently-smart
<22> Athas, z0d thank you
<25> Is it possible to write a macro that does constant folding of forms that are not an argument of the macro? Thus instead of (const-fold-macro <body of code>), you can have <body-of-code> followed by a macro call? I tried searching for "global macro" lisp or "non-local macro" lisp, but I guess I use the wrong terms.
<17> antifuchs: A sufficiently smart compiler would use taht anyway...
<17> fasta: A macro is a comile-time code transformation, replacing the macro form with its expansion. This, alas, means that it can only constant-fold "within its arguments", as it were.
<7> antifuchs: A sufficiently smart compiler would do what you mean.
<25> ingvar: so, if you want to do something "non-local" you need to use setq? For example it's impossible to ask for a current representation of every defined form in your program?
<25> A sufficiently smart compiler would sit at the beach in the sun.
Return to
#lisp
or
Go to some related
logs:
gaim HTTP method
#perl
ubuntu ndiswarper
vector linux root login
#perl
!!! Unable to build DRM modules.
#lisp
problem pyhook encoding
quak 3 gentoo
kubuntu X nolisten
|
|