freenode/#sbcl - IRC Chatlog
Search
16:20:48
dougk_
there may be a silver lining to "no size function for object" bug that randomly occurs in threads.impure.lisp
16:21:30
dougk_
I think it happens when we're looking at a page with pinned code, but other objects were on the page before the pinned code, and those objects were evacuated from the page.
16:23:19
dougk_
basically gc_search_space can't step through the page because it has some forwarding pointers on it. https://sourceforge.net/p/sbcl/sbcl/ci/a33095b0 removed the ability to step over forwarding pointers
18:25:07
knobo
I get [pid 10976] write(2, "Memory fault at 0x3015b1b4 (pc=0"..., 63Memory fault at 0x3015b1b4 (pc=0x1005b19da7, sp=0x7ffff64a6928)) = 63
19:43:50
attila_lendvai1
knobo: the bugs introduced by safety 0 and ffi are often random in nature because they often depend on the actual memory layout
19:45:06
attila_lendvai1
so, what's the plan with the ASDF issue? ASDF fix, or SBCL rollback? or SBCL marches forward to implement some extension for base-string writing?
20:04:36
phoe
I made a build today on another machine, "sh make.sh --fancy --arch=x86"; afterwards, ./run-sbcl.sh had ASDF and all contribs.
20:33:40
phoe
The contribs are build and compiled, but they are not included in the final Lisp image.
20:56:09
attila_lendvai1
stassats: yeah, but it's not an appropriate change in the freeze, or is it?
21:00:06
attila_lendvai1
and the mentioned ASDF fix of adding *read-eval* t also smells... so, not sure what would be an easy way out. I reverted back to the previous sbcl version to be able to continue work.
21:15:39
dougk_
which change causes the problem with ASDF - "string output stream" change or "pathname namestring memoization" changes ?