libera/#commonlisp - IRC Chatlog
Search
12:59:14
bollu
Why do we have two types of quotations? '(1 2 3) and `(1 2 3)? From what I can tell, unquoting (,x) is only allowed within a backquote (`) and not within regular quotes ('). I Why does just backquotes not suffice?
13:01:27
jackdaniel
backquote is meant for templating and it is used by the reader; quote is a special form (quote foo) and it prevents evaluation
13:02:19
moon-child
jackdaniel: technically quote is a special _operator_, and it is (quote foo) that is a special form :)
13:03:53
jackdaniel
they should be treated as orthogonal features acting on a different level of abstraction
13:04:12
moon-child
there's a paper on the topic that iirc constructs something useful (though without nesting) in a page or so of code
13:05:41
jackdaniel
does the standard say that it creates such a thing? (putting aside cltl2 appendix that provides example implementation that does that)
13:06:17
moon-child
'An implementation is free to interpret a backquoted form F1 as any form F2 that, when evaluated, will produce a result that is the same under equal as the result implied by the above definition, provided that the side-effect behavior of the substitute form F2 is also consistent with the description given above'
13:08:20
beach
jackdaniel: It was a way of answering the question "Or are they both primitives?" by giving some better terminology.
13:08:34
jackdaniel
I had quite a headache persuading someone in the past that the implementation is not obligated to create an intermediate form
13:09:07
beach
jackdaniel: That I can understand. The reader macro would be fairly complicated then.
13:09:49
jackdaniel
the reader macro would be fairly complicated when it uses or when it doesn't use the intermediate macro form?
13:09:54
moon-child
https://3e8.org/pub/scheme/doc/Quasiquotation%20in%20Lisp%20(Bawden).pdf pp26-27 here
13:10:43
beach
jackdaniel: The latter. In the first case, it is straightforward. The complication is then in the macro definition of the macro call that it generates.
13:10:47
phantomics
Could someone remind me of the repository that has the suggested CL standard additions?
13:11:18
jackdaniel
I see. fwiw SBCL uses an intermediate form (after cltl2 appendix) while ECL doesn't
13:11:56
beach
phantomics: WSCL is not for additions to the standard. It is for specifying currently unspecified behavior in a way that is consistent with what major implementations already do.
13:14:20
pjb
bollu: you don't need quote either. Instead of writing (list 'a 'b 'c) you can write (list (intern "A") (intern "B") (intern "C")).
13:14:46
beach
phantomics: There is not much that the community would agree upon in terms of suggested additions, so WSCL is meant to be nearly completely uncontroversial.
13:15:53
_death
moon-child: nice, I should read it sometime.. however, even short papers (and code) can take nontrivial time and effort to implement
13:16:18
phantomics
beach: right, the addition I'm proposing will address the definition of an alphanumeric character, which is currently somewhat inconsistent
13:23:18
moon-child
so I do not find fault with that part, unless you limit yourself to _only_ symbols and lists
13:25:18
semz
but if you'd put the object into a form and evaled that, the object would self-evaluate
13:26:14
phoe
the question is whether this is a value that is already read or if it is text meant to be turned into a Lisp value by the Lisp reader
13:27:05
jackdaniel
since we're talking about "objects" not "source code text" there is no such a question, that's what I've confused.
13:27:13
semz
we clearly need LISPRC so we can pass objects to the channel instead of serializing them to text
13:40:17
_death
semz: your "only exceptions" question reminded me of sb-kernel:make-unbound-marker... (eval (sb-kernel:make-unbound-marker)) => #<unbound> hmm ok.. but then, (eval *) => ; Evaluation aborted :)
13:42:01
semz
_death: (eval (sb-kernel:make-unbound-marker)) works for me, as does (let ((x (sb-kernel:make-unbound-marker))) (eval x))
20:08:29
foxfromabyss
1) is it possible to extend existing methods(?), such as `+` or `eq` for new classes, without jumping through too many hoops?
20:08:29
foxfromabyss
2) let's say I have 2 classes. Class A and Class B. Class B has a field with value of Class A. I have written a comparator for Class A. Is it possible to piggyback on that comparator for `sort`?
20:11:13
pjb
foxfromabyss: yes: (shadow '+) (defgeneric + (a b)) (defmethod + (a b) (cl:+ a b )) (defmethod + ((a string) (b string)) (+ (parse-integer a) (parse-integer b)))
20:12:42
foxfromabyss
too many hoops is just me being lazy, and shadowing feels like smth that would break a lot of stuff, but maybe not
20:13:54
foxfromabyss
unrelated, is there an example somewhere of how people use SLIME/SLY? So far i've been just pasting stuff from the source file and testing it there, but it feels like I am underutilizing a lot of features
20:18:51
contrapunctus
foxfromabyss: `M-x slime` -> `(ql:quickload :my-project)` -> edit code in buffer -> `M-x slime-eval-defun`
20:20:38
pjb
foxfromabyss: you can avoid shadowing, by using symbols with a different name: (defgeneric plus (a b)) …
20:25:45
frodef
Hi all, is there some package that provides a compatibility layer for the MOP? Such that I can call e.g. sb-mop:class-slots without relying explicitly on sbcl/sb-mop ?
20:27:00
jackdaniel
c2cl supplements implementation-specified operators like defmethod with wrappers when the implementation doesn't implement fully the mop protocol and that can be fixed with a wrapper
22:01:29
yottabyte
why would one use flat instead of let? you can define functions (lambdas) with let, no?
22:03:03
semz
If you define a function with (let ((f (lambda ...))) ...) you have to call it with (funcall f ...) rather than being able to use (f ...)