freenode/lisp - IRC Chatlog

15:28:37 jackdaniel in SBCL when you cross compile you need to have two hosts available

15:28:49 jackdaniel one with host architecture and one with target architecture (during the building process)

15:29:26 jackdaniel so it is way easier to grab some already ported implementation and boostrap SBCL directly on the host. but yeah, GCC is for instance bootstrapped from GCC, not from assembler

15:29:38 jackdaniel same tactic is valid for Common Lisp

15:38:08 fiddlerwoaroof butterthebuddha: you have to be a bit careful when redefining a constant because it won't necessarily update all the references to that constant

15:39:00 fiddlerwoaroof The compiler will, on occasion, do things like inline a constant's value and then those references won't necessarily get the new value.

15:40:55 fiddlerwoaroof So, the "redefine constant" restart can lead to an inconsistent application state: it's probably better to use reader conditionals to use defparameter in your development environment and defconstant in production, or something like that.

15:41:25 axion I think alexandria:define-constant solves some of this

15:45:41 Bike what define-constant does is avoid trying to redefine the constant if it meets the test with the original value.

15:46:20 Bike so if you do (defconstant +foo+ (list 1 2 3 4) :test 'equal) twice, it defines the constant normally the first time, then the second time ignores the new list.

15:48:56 pjb random-nick: yes, clisp has long-floats (and short-floats) that sbcl doesn't have.

15:49:10 pjb random-nick: also, you can set the precision of long-floats to whatever you want.

15:49:39 Bike sorry, if you do define-constant ... twice.

15:49:51 random-nick pjb: ok, that's useful

16:55:55 dlowe fwiw, I pretty much never use constants ever in CL.

17:01:19 jasom The real sollution for string constants is: (define-symbol-macro +foo+ "some string") /s

17:02:24 pjb (define-symbol-macro +foo+ "some string") (let ((+foo+ "some other string")) +foo+) #| --> "some other string" |#

17:02:38 pjb Do not name it +foo+!

17:05:43 jasom pjb: I think that LET will be a style-warning on sbcl at least.

17:07:15 Bike shadowing symbol macros should be silent, i would think

17:07:35 jasom Bike: not if they have earmuffs

17:07:48 Bike oh, i guess

17:08:10 jasom IIRC sbcl will give you a warning for (let (*foo*) ) or (let (+foo+)) because of the earmuffs

17:08:17 jasom s/warning/style-warning

17:15:06 dlowe probably not for *foo* since that's idiomatic and useful

17:15:36 dlowe I can't think of a good reason to locally bind a +foo+ variable, though

17:18:59 Bike for *foo* when the binding is lexical.

17:25:15 basket dlowe: Lexically binding a constant is illegal

17:27:51 dlowe basket: yes, but that doesn't actually stop you from doing (defparameter +foo+ 5)

17:28:21 dlowe and in that case, it would be legal, but still inadvisable, since it's clearly intended to be constant-ish

17:35:33 Th30n` ** NICK Th30n

17:46:44 _user does anyone know of a common lisp docstring documentation generator that works with org-mode markup, other than this one: https://github.com/jphmrst/cl-org-sampler

17:51:07 malice shka:?

18:33:46 butterthebuddha "The ~t directive is for tabulating. The ~10t tells FORMAT to emit enough spaces to move to the tenth column before processing the next ~a. A ~t doesn't consume any arguments."

18:34:05 butterthebuddha What if the first "~a" is longer than 10 columns?

18:34:53 Bike what first ~a?

18:35:54 butterthebuddha For example, "(format t "~a10t~a" "gobblesmackfoobar" "foobar")

18:36:49 butterthebuddha (format t "~a~10t~a" "gobblesmackfoobar" "foobar") rather

18:39:10 malice Well, obviously you can't go back.

18:39:43 malice I believe it inserts at least 1 space, and at most enough to get into the nth column

18:40:09 malice So, the result would be "gobblesmackfoobar foobar", I guess.

20:19:42 malice How do I run slime-inspector on returned value in REPL?

20:20:01 malice I have something like #<someobject>, but when I press C-c I in there I get error about illegal sharp macro character

20:20:08 flip214 malice: inspect *

20:20:13 flip214 that's the last result

20:20:20 flip214 and ** would be the previous-to-last

20:20:24 flip214 and there's *** too

20:20:32 _death C-c I * <RET>

20:21:05 scymtym C-c C-v TAB when the cursor is on the presentation

20:21:07 anticrisis malice: try also C-c C-v TAB whith your cursor

20:21:19 malice flip214: but that would be inspector

20:21:29 malice I want slime's inspector, so I can move around with shortcuts and cursor

20:21:46 malice thanks _death

20:22:02 malice and thanks scymtym and anticrisis

20:22:04 malice both approaches work

20:24:30 axion If you use Sly, nearly everything is a button that can easily be inspected.

20:25:52 flip214 malice: I meant to use the slime inspector on *

20:44:41 gnuhurd hi, if I have a function that setf's a variable, and I want to return the variable at the end, can I use eval? will it pose any risks? if not, what should I do to return it?

20:46:26 gnuhurd this is how my function looks currently: https://p.teknik.io/uXWw8

20:46:58 Xach gnuhurd: you can just end with mft-name. no need for eval.

20:47:34 Xach the value of the last form evaluated in the function is what gets returned. and evaluation is "automatic" in this context.

20:47:48 gnuhurd thank you

20:50:39 jasom are there any decent refactoring tools for lisp (e.g. rename a lexical binding)? I currently make do with find/replace but that seems slightly awkward for some code.

20:51:17 Xach jasom: there was redshank, but i never tried it and i think it was pretty limited in scope.

20:51:26 jasom other things that would be nice to automate is: turn this block into a defun, with parameters named the same as all externally-bound values referenced in the block

20:52:22 jasom (let (foo) (let (bar) (baz foo bar))) <-- e.g. it would be nice to be able to turn the (let (bar) ...) into (defun _CURSOR_ (foo) (baz foo bar)) at the toplevel somewhere

21:13:19 malice flip214:then I didn't understand you, sorry