freenode/#sbcl - IRC Chatlog
Search
6:32:20
flip214
http://paste.debian.net/1100992 ... that my filename is bad (because of non-escaped []) is okay, but why the FORMAT-ERROR?
8:42:26
scymtym
flip214: that's one of the two or three bad ones. anything after 4424f6780 should fix the problem
8:49:27
scymtym
flip214: sure. sorry for breaking it in the first place. i guess that's how clean ups go sometimes
8:51:21
flip214
scymtym: at least there's another test that can be added to avoid similar problems in the future...
8:52:40
scymtym
flip214: right, i added test cases for the restarts. should have done the same for the file errors
11:13:27
stassats`
no-defun-allowed: what do you mean, replace the pointer? it just unprotects the page and marks it as potentially-containing-pointers-to-new-space
11:15:16
no-defun-allowed
Don't worry about it, I had a talk with some Un*x weenies and it doesn't seem trivial to do what I was thinking of, and I don't exactly know what I want either.
11:16:23
no-defun-allowed
I was just looking if anyone had experience with implementing read/write barriers, since libsigsegv has no documentation and there isn't much documentation on using stuff like mprotect + sigaction to implement them either.
11:17:55
stassats
if you don't unprotect the page you'll quickly kill performance if it keeps pounding the same page in a loop
11:20:23
no-defun-allowed
So if I was reading off quite a lot of data in oldspace in a Baker collector, then a software barrier be better?
11:25:34
no-defun-allowed
No problem then, again I should ask somewhere else if it isn't an SBCL thing.
16:36:31
dougk_
fitzsim: yes, the error when building sb-sprof on ppc64be is repeatable until someone fixes it.