libera/#clasp - IRC Chatlog
Search
14:39:14
yitzi
Bike: I've gotten the scaper logical paths to work. https://plaster.tymoon.eu/view/3254
14:46:32
yitzi
Just for references, here is the translation table for SYS https://plaster.tymoon.eu/view/3255
14:48:03
yitzi
119 elements, but it probably really doesn't matter since it will only be used if someone does translate-logical-pathname on a source reference. Otherwise, won't be used.
15:00:07
yitzi
I am also trying to improve the output from clasp-builder. Make it a little easier on the eyes.
15:01:34
yitzi
And I have an idea to use the time that it takes to load each file in an image to sort the files in descending order before compilation. Might speed up compilation of of a[bce]clasp a bit.
15:20:16
Bike
i built clasp from clean and i hit a problem loading/compiling trivial-features/src/tf-sbcl.lisp because sb-alien does not exist.
15:26:12
yitzi
Bike: I should have put a note in the PR about that. You need to delete build, dependencies, and src/lisp/kernel/contrib
16:30:30
Bike
not sure if i did something wrong, but source locations don't seem to be working right? M-. gives me that characterp error. the source filename for core:file-column is given as #p"sys:src;core;lispStream.cc", which seems like it ought to be uppercase
16:32:07
Bike
TAGS also being weird... i took the etags thing out of my config a while back, but it's still failing to find pretty uncomplicated definitions
16:36:56
yitzi
I didn't change anything for source locations there. That is a different branch. The scraper source reference like #P"sys:src;core;lispStream.cc" won't work because of the mangled case.
16:43:48
yitzi
The logical pathname stuff for source locations I have mentioned is in `yitzchak_logical-kernel-paths`. In main, the source references that come from the scraper are probably broken logical paths. The rest are stils absolute paths.
16:45:32
yitzi
Yes, well....that would be the easy way to say it. I was going for convoluted and obscure.
16:46:37
yitzi
I still need to test the logical source references for an installed clasp and for snapshots.
17:05:33
Bike
https://github.com/clasp-developers/clasp/issues/1348 minor: not sure if this is a bug
17:19:32
yitzi
I wouldn't be surprised that clisp's printer (or pretty printer) behaves differently from other implementations. As I recall even the pretty printer is not derived from the original XP code and has a lot implementation in C code.
17:21:40
Bike
it's not actually a pretty printer thing. i replicated this behavior in clisp by giving it a longer repl prompt.
18:53:24
kpoeck
the asdf self test complain - correctly that we have code-weight.asd in profiler, once https://github.com/clasp-developers/clasp/blob/main/src/profiler/code-weight.asd and once in https://github.com/clasp-developers/clasp/blob/main/src/profiler/code-weight/code-weight.asd. Surprisingly the file code-weight.lisp is also different in the two places.
18:53:24
kpoeck
And file there is package definition n one asd that doesn't seem to be used. The description of this system is: Compile-file files with :output-info t and generate .ll files and then pass a list of .ll files to the analyze-code function.
18:54:29
kpoeck
Can't see that analyze-code s called anywhere so probably just an utility. Any preference on how to clean that up? If not I would just do a PR with my best educated guess
19:15:03
kpoeck
yes, and probably code-weight.lisp for consistency. And if we create a package #:code-weight, we might as well use it in code-weight.lisp