libera/#clasp - IRC Chatlog
Search
6:33:20
pfd
I just tried a (buildmode (:bitcode)) build with (:extensions (:cando :seqan-clasp)) and it failed at the same attempting Linking of:
6:33:21
pfd
Perhaps this is not just me. Maybe it's a current failpoint for others too. I say this only because I thought a noticed a discussion
6:33:22
pfd
earlier yesterday regarding this boehmprecise gc; or did I notice this discussion with rotateq in another chat group #phantomics (APRIL APL compiler to CL).
10:57:28
pfd
../src/gctools/snapshotSaveLoad.cc:2117: warning: the use of `tmpnam' is dangerous, better use `mkstemp'
10:57:29
pfd
../src/gctools/snapshotSaveLoad.cc:2412: warning: the use of `tmpnam' is dangerous, better use `mkstemp'
10:57:30
pfd
../include/clasp/gctools/snapshotSaveLoad.h:254: error: undefined reference to 'dladdr'
12:05:03
yitzi
Bike: Looks the clasp_gc.sif file got overwritten by static analyzer when cando was loaded. Reran the static analyzer on ci branch and it build sucessfully.
13:01:13
pfd
CLASP_QUICKLISP_DIRECTORY=../dependencies/quicklisp-client/ boehmprecise/iclasp-boehmprecise --no-rc --non-interactive --load snapshot.lisp -- boehmprecise/dcando-boehmprecise
13:01:14
pfd
Should run ./koga --ldlibs=-ldl again, followed by ninja -C build ? Or should I start all over with a different config; maybe with the default of :faso instead of :bitcode ...?
13:08:29
yitzi
pfd: Are you trying to use cando? If you just need Lisp then you should probably configure with out it.
13:10:18
pfd
OK. I'll start again without cando and seqan... I thought I would try adding them, thinking, somehow what they might pull down might help.
13:17:43
yitzi
You can get the ldlibs option I mentioned earlier with this config.sexp https://plaster.tymoon.eu/view/3128#3128
13:27:38
yitzi
replace "about" in my previous sentence with "able." Its silly shorthand based on sed.
13:30:11
yitzi
Bike: I think we should just merge https://github.com/clasp-developers/clasp/pull/1304 ... it has errors, but only because we have some failing ANSI tests.
13:33:58
yitzi
I have not. Bike had fixed several ones, I believe. I can do a PR to kpoeck if we want to move failing into expected if you want.
13:34:41
Bike
i think we should do that. then it should at least be obvious if a change causes more failures.
13:36:28
Bike
and yes, i think i fixed several, but i didn't actually rerun the test framework (just checked some individually)
13:40:41
yitzi
Crap, just got the SBCL 2.2.4 problem on the MacOS test. They must have just updated homebrew.
13:41:50
drmeister
I can run the static analyzer with 10 threads and get about 700% CPU utilization. It takes 7.5 min.
13:43:27
pfd
Oh, I'm using SBCL 2.2.4 as well! Maybe that was my problem. Maybe I should step back to 2.2.3 if my current build fails.
13:43:54
yitzi
I know about to avoid it, thats about all. I think it has to do with source code line tracking with `--script`