freenode/#clasp - IRC Chatlog
Search
15:26:39
drmeister
beach: Oh - this is what this is about? Do what you need to do in Cleavir - we will fix the implementation on this end.
15:27:09
drmeister
Don't twist the design of Cleavir to meet the peculiar limitations that Clasp currently has.
15:28:34
beach
Bike: I would need a mechanism for creating environments, separate from the existing one.
15:28:48
Bike
I thought the idea was that we would have it as ((lambda (...) (lambda (x) ...)) ...), and then we'd do the cell conversion, and then we could still eliminate the call, and just put the make-cells where the call was.
15:30:16
Bike
Since we don't like... create environments, per se, in HIR, just cells, which are passed to ENCLOSE like arguments.
15:31:31
beach
It all requires the better control-flow and data-flow analysis that we have discussed several times.
15:33:31
drmeister
Bike: I have an issue that just came up with fastgf. I'm running the acid test and an error was triggered:
15:33:46
drmeister
((METHOD CLOS:SLOT-MAKUNBOUND-USING-CLASS (BUILT-IN-CLASS T T)) #<VA-LIST: (#<The STANDARD-CLASS STANDARD-CLASS> #<UNPRINTABLE> 'NIL #<The STANDARD-CLASS STANDARD-CLASS> #<The STANDARD-CLASS CLASS> #<STANDARD-EFFECTIVE-SLOT-DEFINITION CLOS::VALID-INITARGS> )
16:20:26
drmeister
The problem may be with the effective method functions calculated during satiation
17:39:02
drmeister
Bike: if Shiho is there - can you ask her to read the clhs page on make-array, aref, svref and so on? She isn't online at the moment
18:52:43
drmeister
This is driving me nuts. If I build bclasp - it fails because slot-makunbound-using-class invokes the wrong method.
18:54:26
drmeister
(slot-makunbound-using-class #<The STANDARD-CLASS STANDARD-CLASS> #<The STANDARD-CLASS CLASS> #<STANDARD-EFFECTIVE-SLOT-DEFINITION CLOS::VALID-INITARGS>) is invoking the (defmethod slot-makunbound-using-class ((class built-in-class) self slotd) ...)
19:40:03
Serenitty[m]
I finished compiling Clasp on the slow computer, too. So I think it's fair to say that Clasp builds fine on Arch, now.
19:55:13
Serenitty[m]
So, I've generated the bytecode file from the C++ code in the demo, but trying to load the Lisp file results in a crash. I guess Linux compatibility still needs work.
20:35:05
Kevslinger
ACTION likes Common Lisp much much better than the languages being taught in University
20:42:25
Serenitty[m]
Huh, so I went back to open up SBCL, since I felt like using it for doing some Lisp development, but it isn't working anymore. I'm getting the error that it can't find NIL/sbcl.core. That sucks.
20:43:54
Serenitty[m]
Hmm, no. Where should it be set? I have the package from the repositories installed.
20:46:39
Serenitty[m]
Huh. It seems that Emacs is trying to set it whenever I open SLIME. Maybe it has some stuff left over from Clasp. I'll try wiping my cache or whatever there is in my .emacs.d.
20:49:49
Serenitty[m]
Oh no, it's doing it again. Okay, what was your advice? What file should I be grepping?
0:59:26
drmeister
Hmmm - for some reason that problematic generic function is completely bypassing dispatch and going straight to call the effective method.
2:04:16
drmeister
shared-initialize is being called on generic-functions over and over and over again - should that be the case?