freenode/#lisp - IRC Chatlog
Search
2:28:59
axion
Does anyone have any resources on how to use CI/git hooks/asdf test-op to run test suites on commit etc? This idea seems very interesting to me for a large math project of mine, but I have only briefly heard of it before.
6:12:45
impaktor
Where/how is $SBCL_HOME set? Mine is just NIL. (installed through Arch Linux package repository).
7:03:22
_death
I don't understand the issue they think they solve, but the other day they committed something that may "fix" your issue (though I've not tried it.. I think I just removed the putenv call) so you may want to try it
8:14:58
lieven
since threads aren't mentioned in the standard, surely this is implementation specific?
8:16:40
jackdaniel
lieven: in principle you are right, but in practice it costs too much to make pop atomic by default
8:19:52
_death
knobo: it means the assumption that normally a resource is free for use and otherwise is the exceptional case
8:21:21
pjb
And what atomicity do you want in (pop (aref (give-me-some-vector foo) (incf (elt indices (incf j))) (random 3))) ?
8:22:31
pjb
Perhaps you need one lock per array slot, or per indices, or for the whole expressions; the implementation cannot know: it depends on the semantics of your application.
8:32:39
axion
_death: thanks. I am trying to use 'prove' to run tests on travis-ci, but for some reason failed tests are still exiting cleanly with status code 0 and even though the tests are shown as failed, travis reports everything as passed. anyone know what could be causing this?
9:14:58
aerique
If I find an issue with a library (clsql in this case) loaded through Quicklisp it is not obvious to me from which repo Quicklisp got that library. How do I find that out?
9:20:13
aerique
ah: https://github.com/quicklisp/quicklisp-projects/blob/master/projects/clsql/source.txt
11:05:17
axion
How can I retrieve how an object is printed as a string? I wrote a custom print-unreadable-object form to print an object, and I want to write a test in my test suite to make sure it is exactly equivalent to a particular string. is this feasible?
11:12:37
axion
phoe: this will be a multi-line matrix in a specific order not like the memory layout of the array
11:17:12
phoe
"By default, a structure of type S is printed using #S syntax. This behavior can be customized by (...) or by writing a print-object method that is specialized for objects of type S."
11:18:31
phoe
flip214: yes, but the structures are compiled into vectors. Once they are created, Lisp does not see the instances of that structure as structs, but as vectors.
11:19:48
flip214
phoe: but 22.1.3.12 Printing Structures doesn't talk about the :type of structures... I read that as if it should work for any internal representation
11:20:19
axion
It very well may. I will have to try. The point of this structure is to have native arrays with struct slot accessors
11:20:46
axion
Thanks! Though, how can I retrieve the printed representation of an arbitrary instance in my test suite?
11:21:41
flip214
perhaps the first slot of that "struct" can contain some gensym, which gets recogniced by some user-defined PRINT-OBJECT method or so?!
11:22:12
axion
and by the way, the struct slot accessors on the array compile with the same disassmebly, on at least sbcl. this is nice because this is an efficient math library and makes the code very concise and easy to understand
11:23:46
phoe
flip214: it cannot contain gensyms if the array is specialized, like here - for single-floats.
11:24:11
flip214
phoe: well, then just stuff some "special" single-float value in there... the NaN space is big enough, innit?
11:24:54
flip214
I understand that a struct will need a bit more space... but I don't think there's a performance difference.
11:25:19
axion
much, on at least sbcl. with the type explicitly as a simple-array single-float 3, the assembly is much better
11:26:14
axion
well yeah, that is my friend's library that i linked. i am doing a similar fashion with singles.
11:27:51
axion
i didn't really want to get into the why i did this thing again...it seems to be a monthly question from this channel. we have done extensive tests over the last 3 years regarding this. that's why :)
11:30:07
axion
you'll have to ask psilord on the low level specifics anyway. he is mostly just in #lispgames though