freenode/#clasp - IRC Chatlog
Search
16:01:59
drmeister
I've learned over the last few days that genetic algorithms are amazingly powerful. It doesn't matter how effed up your scoring function is - it will optimize it.
16:12:24
Bike
if i have a file with line 1 (defun foo (), line 2 (error "test")), and compile it with source-debug-offset 4, the DILocations in the ll are on line 5 and 6
16:15:39
drmeister
Or add 1 - but probably subtract 1 Miss. Melinger in second grade was right when she said that arithmetic would be valuable to me later in life.
16:33:05
Bike
compile-file-source-pos-info returns pretty m uch the same thing for C-c C-c and a regular compile-file
17:39:47
Bike
in compile-file we have with-debug-info-generator and we give it the compile file truename even if we don't actually want that. o-k.
18:29:10
Bike
we may want to reevaluate if we actually want the file part of a source pos info to be ignored, though
19:37:28
drmeister
Rectangular periodic boundary conditions (wrap around map) with hexagonal tiles is easiest represented by offset coordinate system https://usercontent.irccloud-cdn.com/file/JUlwOmyQ/1569353791.JPG
19:45:12
Bike
just pushed the source debug lineno thing. here's my slime/swank/clasp.lisp to use it - basically just passes a :source-debug-lineno to compile-file. http://ix.io/1WLn
20:35:43
kpoeck
https://gitlab.common-lisp.net/alexandria/alexandria/blob/master/functions.lisp#L157
21:32:40
scymtym
::notity kpoeck: i am of course considering relaxing the constraint. but i must say that i'm not too happy how relaxing it is treated as a no-brainer, apparently giving little thought to the question which behavior is conforming
21:32:40
Colleen
Unknown command. Possible matches: notify, 8, set, mop, get, login, roll, say, uptime, grant,
21:32:48
scymtym
::notify kpoeck: i am of course considering relaxing the constraint. but i must say that i'm not too happy how relaxing it is treated as a no-brainer, apparently giving little thought to the question which behavior is conforming
21:33:10
scymtym
::notify kpoeck: also note that (handler-bind ((eclector.reader:unquote-in-invalid-context #'eclector.reader:recover)) (read ...)) will be (that is almost already the case but not quite) enough to make this and related issues go away
21:50:56
drmeister
scymtym: I'm missing the subtlety - (using an example you wrote in the logs)... I thought if #' is syntactic sugar for (FUNCTION ...) then `#'(lambda () ,(random 10)) should be treated like `(FUNCTION (lambda () ,(random 10))) - which I thought should be allowed.
21:55:43
scymtym
drmeister: i think it should be allowed as well. what i'm trying to find out is whether there is evidence in the spec that it actually is allowed
22:05:42
drmeister
I can't find a discussion that I had with fare in the #lisp logs but I saved it for this verry moment.
22:06:41
drmeister
It doesn't answer the question but near the bottom Fare says that the lack of information about how reader macros should deal with quasiquote and , and @, is a bug in the specc.
22:21:42
scymtym
i agree with fare that a protocol for user-extensible quasiquotation would have been good
22:41:25
Bike
drmeister: the problem is that the spec only defines how `( and `#( work. `#' is not defined. There is a fairly obvious interpretation of course, but strictly speaking it's not standard.
22:46:08
scymtym
i think in some cases it's just how the respective implementation of the reader macro happens to behave
22:47:55
scymtym
there are some deliberate decisions: clisp says "unquotes may not occur in structures"
22:49:19
Bike
i think with structures it's related to the fact with the usual way it goes you'd have to iterate through structure slots, which is kind of messy
22:51:00
scymtym
sbcl has this very specific one: "Comma inside a backquoted array (not a list or general vector.)"
22:54:44
scymtym
maybe i should make WITH-FORBIDDEN-QUASIQUOTATION client-extensible so each client can choose what to allow