libera/#clim - IRC Chatlog
Search
14:14:26
scymtym
strange design decision. EQUALP should be almost as efficient in the unmodified case but the with sane semantics
14:15:39
jackdaniel
well, that won't work, because the new value is setf (not copied) to the cached version
14:19:08
scymtym
i haven't looked at the code but shouldn't the contents of the vector be copied when the new clip mask is different? copying can be done without consing if the cached vector is kept around and modified
14:20:59
jackdaniel
it would need to be adjustable (there may be more than 4 values) - otherwise than that - sure. that's a solution I'n going to submit
14:24:23
scymtym
i think i lean towards reallocate since it saves the memory and computation for the extra indirection and is easy to implement as (if length-is-different (make-array … :initial-contents new-clip-mask) (replace …))
14:51:16
jackdaniel
if you want to take look then here's nothing: https://github.com/sharplispers/clx/pull/202
15:13:41
scymtym
i can't conveniently log into github right now so i will comment here: couldn't https://github.com/sharplispers/clx/pull/202/commits/21108d4ff60ff994a3154317fb4117bb847ea25b#diff-46941f928f5a36231eccb808aa942c0573b4c4c3e9b9c5c608fce9f64bcfcdf2R606 use CLIP-LENGTH?
15:15:19
scymtym
since the issue is a bit subtle, i think it would be good to explain the problem and solution in the commit message (i don't trust github issue references as a long-term source of information)
15:16:06
scymtym
maybe a one or two sentence comment here https://github.com/sharplispers/clx/pull/202/commits/21108d4ff60ff994a3154317fb4117bb847ea25b#diff-46941f928f5a36231eccb808aa942c0573b4c4c3e9b9c5c608fce9f64bcfcdf2R599 to explain broadly what the following lines do?
15:17:00
scymtym
ACTION is a bit paranoid about not understanding in the future why changes were made
15:40:19
jackdaniel
it is good that I have a dinky laptop because some performance issues (with the new xrender implementation) are not visible on the other machine