freenode/#lisp - IRC Chatlog
Search
1:16:11
fiddlerwoaroof
The standard does have a provision that says something like "on a parse error, a conforming parse error may signal an error" or something like that.
2:07:39
verisimilitude
Who knows, Duns_Scrotus, but we know they're just looking at Chrome for it, now.
2:08:12
verisimilitude
In any case, would I be correct in thinking no current Common Lisp HTML parser takes this approach and has customizable parse error behavior?
2:10:14
Duns_Scrotus
it's not really "heuristics" it's a bunch of huge explicitly described state machines https://html.spec.whatwg.org/multipage/parsing.html#overview-of-the-parsing-model
2:22:48
verisimilitude
My idea was to have a simple parser that would ideally be fast and to only incur the complexity of heuristics if needed and optionally.
3:22:38
ym
M-. in last (quicklisp) slime/swank fails with "cond: Error: end of file on #<SB-IMPL::STRING-INPUT-STREAM {...}>"
4:51:02
fiddlerwoaroof
emaczen: https://github.com/fiddlerwoaroof/objc-lisp-bridge/blob/master/demo-app.lisp#L130
4:51:31
fiddlerwoaroof
https://github.com/fiddlerwoaroof/objc-lisp-bridge/blob/master/demo-app.lisp#L125
4:55:55
emaczen
If you can add methods at runtime, can't you then just call performSelectorOnMainThread:withObject:waitUntilDone:
5:00:16
emaczen
fiddlerwoaroof: What exactly is the superclass parameter to objc_allocateClassPair ?
5:01:13
emaczen
the documentation is just confusing to me since it seems to say that it is the metaclass of the class you are creating, but then why name the parameter super_class?
7:24:23
LdBeth
<no-defun-allowed "Fair enough."> #'no-defun-allowed: statistics showed CDR coding doesn’t reduce space used in practice
7:25:03
no-defun-allowed
That's very odd then. I read LispM compactors would try to create CDR coded lists.
7:28:52
LdBeth
With CDR coding they cannot just copy the list, because the reference show be kept anyway, so after all fragmentation would be cause
8:53:39
phoe
If it's a term already coined by R, then it sounds like naming it a data frame will be good enough.
9:01:17
shka_
phoe: anyway, i want to implement data-frames and eventually stuff like that https://github.com/rstudio/cheatsheets/blob/master/data-transformation.pdf but only better
9:22:02
LdBeth
beach: I maintained a copy of CLFSWM, with a LispM Flavor package, I have privately kept a fork of CLWEB (I’ve send patches to the original author but seems he’s to busy to merge them), that’s all.
9:34:44
no-defun-allowed
It's not the software's fault, but when anyone else needs it, there's always a conflict and it ends in tears.
9:36:18
no-defun-allowed
jackdaniel: woodworking generally ends in tears and bloodshed quicker than programming, at least
9:38:04
LdBeth
phoe: that’s why people have to learn typologia and place line breaks by hand. Before an AI is made for working on this.
9:39:41
no-defun-allowed
though it's more fun than overfull buffer (badness [Segmentation fault, core dumped]
10:15:53
jackdaniel
I don't know what a trivial-configuration-parser is, but there is no magic, ssl is bound to :tt
10:52:19
jackdaniel
jimmyjoe: I've heard this "advice" over and over again from people involved with security: never write crypto library because it is hard (imo it is a dumb advice)
10:52:42
jimmyjoe
what about other interpreters like python,ruby,etc - these libraries claim to be secure, and widely used.
10:52:59
jackdaniel
but this attitude implies, that *any* crypto library written by a not enterprise is not secure (as if enterprise software were secure)
10:53:38
lieven
jimmyjoe: secure as an absolute is fairly meaningless. you need to specify use case, attack model etc
10:54:49
lieven
jimmyjoe: some of the more hardened libraries try to achieve input independent code paths to prevent power consumption measurement attacks etc
10:56:32
jackdaniel
if you refere to a recent post by Wimpie Nortje it explicitly talks about threading issues
11:03:28
akater
jimmyjoe: What I meant: If you want security, you generally do not want your program to enter a state not belonging to a (relatively) small set of specified, well-understood states.
11:03:34
akater
The idea of an interactive language with rich runtime goes against this. At the very least, the state space will tend to be as *large* as possible, contrary to what you'd want.
11:07:54
White_Flame
ruby etc call out to C libs to perform the crypto. they're not implemented in the interpreted langauge themselves
11:08:11
White_Flame
and as lieven mentioned, timing attacks and such are the primary concern with non-C written algos
11:08:34
White_Flame
those are written to ensure the same amount of CPU work happens regardless of path taken
11:09:24
akater
Would you say it's a bad idea to assign meaning to objects dependening on whether certain slots are unbound?
11:09:30
akater
E.g., (obj :slot nil) means one thing while (obj :slot #<unbound>) means something else, and is also considered a proper object.