freenode/#clasp - IRC Chatlog
Search
23:55:08
Bike
i don't know if it's a cause, but having an entire special operator broken is not good
0:05:50
Bike
oh, and about esrap working a few days ago - does that mean dev from a few days ago, or shiho's much older dev that she only recently updated?
0:14:30
Bike
i cut down allocate-instance to allocating the memory and setting a few C++ fields, so that's nice
0:34:11
drmeister
You think so? Not the (LET ((EXPR-FUN10693 (LOAD-TIME-VALUE (ESRAP::COMPILE-EXPRESSION 'WILD-ATOM-LINE)))) ...)
0:34:53
Bike
besides, if the compile-expression was hanging, it would hang when you load the fasl, right? not when you actually do the parse.
0:35:05
drmeister
I was thinking the load-time-value was wrong and that was screwing up the funcall - but ok.
0:36:37
Bike
so you can do something like (let ((fun (esrap::compile-expression 'wild-atom-line))) (print "done with compile") (funcall fun text 0 (length text))) and see how that goes
0:36:55
drmeister
This (1) stick a print statement in the code (2) compile everything (3) run it - is driving me nuts
2:46:59
drmeister
I've held off debugging issues because I wanted source tracking - and it's in the cst branch.
2:47:31
drmeister
Here I am in the 'dev' branch debugging a nasty problem in compiling esrap expressions trying to read freaking tea leaves to figure out what is wrong.
2:48:12
drmeister
If I can't crack this in the next couple of edit/rebuild cycles I'm merging it into 'cst' and I'll try debugging it there.
2:55:03
Bike
oh byt the way you mentioned having a make-unspecial function- i added a (setf core:specialp) in dev that kind of subsumes that
2:55:42
drmeister
Ok - can you remind me of that tomorrow and I'll remove make-unspecial and change my port of maxima
10:00:21
scymtym
drmeister: i can't pinpoint the problem based on what i have seen in the logs. infinite loops in esrap parsers are usually caused by repetition expressions with child expressions that can succeed without consuming input such as (* (? "a")). also, esrap has a tracing facility that could help (esrap:trace-rule 'RULE :recursive t)
11:27:30
drmeister
I'm going to pull another copy of clasp and cando, merge dev into cst and debug this issue in esrap with the help of source tracing and esrap trace.
11:28:13
drmeister
Something broke in the last couple of days and I need to fix it fast - so that we can get this very important demo running for next Thursday.
11:28:33
drmeister
Or by the end of today I have to start regressing to older versions to find one that works.
11:30:03
drmeister
We can't render molecules because that requires parsing sybyl type rules and that requires esrap .
11:32:06
attila_lendvai
drmeister: do you know how to cherry-pick specific commits? maybe for the demos you should stick to the testing branch. create a branch for the demo and cherry pick selected commits if needed...
11:34:27
drmeister
It looks like it's been in the code since at least Monday - but I've been traveling and giving demos. When I tried to demo the development version it locked up - probably this problem.
11:34:47
drmeister
I dropped back to a docker demo but that precluded me from demonstrating some of the latest features.
11:37:01
attila_lendvai
drmeister: maybe for the demos you should stick to the testing branch? create a branch for the demo and cherry pick selected commits if they are needed for the demo (it's simple with gitk as long as there are no line-diff conflicts).
11:37:24
drmeister
The testing branch doesn't let me demonstrate the latest features that I need to demonstrate by next Thursday
11:39:18
drmeister
esrap:parse is a compiler macro - yes. But the problem happens when it invokes the compiled parser.
11:39:45
attila_lendvai
*if* it's one self-contained commit/feature that's needed for the demo, then cherry-picking can be useful to bring that one change over without all the instability of the other new developments.
11:41:34
drmeister
(ESRAP::EXEC-NEGATION #<CLOSURE-WITH-SLOTS@0x11ce34e88 LAMBDA :ftype :FUNCTION lambda-list: NIL :fptr 0x1261ff610> (NOT (OR #\NEWLINE #\PAGE)) "WILDATOM XX C N O S P" 8 21)
11:42:13
drmeister
It's failing - I'm not sure if that should happen or not. Is it looking for #\NEWLINE or #\PAGE in that substring and not finding it?