freenode/#clasp - IRC Chatlog
Search
1:26:21
drmeister
What's wrong with this construct? It's a handler-bind inside of a handler-case - I'm still wrapping my head around conditions and restarts - does it make sense to have a handler-bind inside of a handler-case like this?
1:27:06
drmeister
Clearly I want to generate the backtrace in the handler-bind serious-condition handler - so that it is generated before the stack is unwound.
1:27:35
drmeister
Should I write it into a closed over variable so that the handler-case simple-condition and serious-condition can print it?
1:30:18
drmeister
I added calls to (trivial-backtrace:print-backtrace-to-stream *error-output*) to the handler-case clauses.
1:31:15
Bike
i'd say you can drop the handler-case and just have a handler-bind that muffles warnings, prints conditions, and for errors prints backtrace
3:55:30
drmeister
balrog: We have a version of clasp that does full source tracking and generates lots of debug information.
3:56:16
drmeister
It builds - but it's not quite ready for prime-time because Bike is still working on getting a new type of inlining working.
3:56:46
drmeister
We have disabled the new type of inlining - and because of this it runs about 2x slower than the 'dev' branch.
5:28:41
karlosz
(clisp-cleavir interface meaning not the stuff in the SICL repository, but all the stuff i defined)
5:33:16
karlosz
drmeister: not quite. i've only tried self compiling the machinery i added. lemme try getting quicklisp to use cleavir to load the other parts of sicl
11:13:56
attila_lendvai
drmeister: if you want control over the waf output then you need to stop using print in wscript. mimic the log.info() or .debug() calls that you find there.
12:06:18
kpoeck
Regarding ansi-tests, I forked the hopefully newest repository from jackdaniel on gitlab and pushed the modifications to run clasp in a branch
12:07:34
kpoeck
Instructions on how to run in https://gitlab.common-lisp.net/kpoeck/ansi-test/wikis/How-to-run-the-tests-for-clasp
12:09:34
kpoeck
Since clasp is relatively slow, I added features to just load parts of the tests, e.g. just sequences, but pushing :ansi-load-sequences to *features*
12:10:35
kpoeck
I did not touch any test from upstream, but deactivated the tests that make clasp exit and defined expected failures
12:13:30
kpoeck
Will try to add a functionality to execute the tests with fork, as clasp does in compiling (but need to figure out, how to communicate the subresults)
12:14:48
kpoeck
On my machine (some fixes not yet merged or not even comitted) I get 750 out of 21724 total tests failed:
12:17:02
kpoeck
obviously will try to clean up and make a pr to jackdaniel repository (after testing my changes with other lisps)