00:22 tokuhiro_ joined 01:23 tokuhiro_ joined 01:57 colomon joined 02:23 leont joined 02:49 tokuhirom_ joined 03:14 tokuhiro_ joined 04:20 tokuhirom_ joined 05:21 tokuhirom_ joined 06:38 Ven joined 07:02 domidumont joined 07:17 Ven joined 07:21 FROGGS joined 07:27 Ven joined 08:16 Ven joined 08:58 pyrimidine joined 09:07 kjs_ joined 09:46 domidumont1 joined 10:01 japhb_ joined, psch_ joined 10:06 Ven joined 10:29 leont joined 11:22 Ven joined 11:37 tokuhir__ joined 12:00 domidumont joined 12:15 tokuhir__ joined 12:16 JimmyZ joined 12:32 Ven joined 13:40 Ven joined 14:32 tokuhiro_ joined 14:46 tokuhirom joined 14:47 TimToady joined 14:55 TimToady joined 15:27 Ven joined 15:31 Ven joined 15:36 TimToady joined 15:56 synbot6 joined
dalek arVM: c06c0cf | (Dagfinn Ilmari Mannsåker)++ | src/strings/ (6 files):
Fix const warnings about pointers to Unicode case table
16:06
arVM: f80c129 | lizmat++ | src/strings/ (6 files):
Merge pull request #285 from ilmari/const-case-changes

Fix const warnings about pointers to Unicode case table
16:06 tokuhiro_ joined 16:09 synbot6 joined
TimToady note: in general, jnthn hates const sprinkling more than the warnings 16:11
16:12 synbot6 joined 16:14 synbot6 joined 16:17 synbot6 joined
lizmat TimToady: but does he hate it because it is tedious work (which would make him happy having it been done already) 16:19
or is it reading the code with const sprinkled which upsets him ?
fwiw, I prefer to see no warnings at all when compiling :-) 16:20
16:20 ilmari joined 16:22 Ven joined
[Coke] the consts tend to cause more trouble than they fix, as I recall. 16:23
ilmari would rather have a compile time error when trying to assign through a pointer to const than a runtime segfault when it tries to write to a read-only mapping 16:24
[Coke] and they're infectious - once you start adding them, you can't stop. :|
ilmari ilmari@garkbit:~/src/moarvm$ ag -w const src/|wc -l 16:25
1205
I'm sure six more don't hurt 16:26
sorry, seven
I'm no proponent of consting things just because you can 16:28
but
a) big tables actually benefit from being const, since they can be mapped read-only and shared between processs
b) "harmless" warnings can mask real problems 16:29
TimToady prior discussion at irclog.perlgeek.de/moarvm/2015-09-23#i_11262041 16:30
leont [Coke]: that is even more so in C++. I tend to start my project const-correct, but adding it later is a pain (and even adding as you go can be annoying) 16:31
ilmari I've seen the trouble consting sprees have caused in perl5, so I'm no fan of retroactively consting stuff like function parameters 16:32
TimToady which you just did 16:33
ilmari not for the sake of it
I meant stuff like "oh, this function doesn't write modify this argument, let's const it" 16:34
this was to fix an actual warning, and deconsting the tables to which they point would be wrong
TimToady well, I'm ambivalent, just trying to stick up for jnthn while he's away :) 16:35
"stick up for" is an odd English idiom... 16:36
I guess it's a protest sign metaphor
16:48 tokuhirom joined
lizmat unless someone stops me, I will revert 16:48
JimmyZ added the const to the global static var, for some good reason.
lizmat: I think it is fine to keep it.
since it is for fixing warnings. 16:49
[Coke] jnthn can revert it when he gets back.
if he wants.
JimmyZ yeah too.
lizmat considers herself stopped 16:51
leont Wait until you add volatile-correctness to the mix, then it becomes fun ;-) 16:52
17:00 tokuhiro_ joined 17:54 FROGGS joined 18:07 vendethiel joined 18:08 FROGGS joined 18:48 domidumont joined 18:49 tokuhirom joined 19:16 tokuhiro_ joined 19:41 pyrimidi_ joined 20:17 vendethiel joined 20:51 colomon joined 21:19 flussenc1 joined 21:46 tokuhiro_ joined 22:47 tokuhiro_ joined 23:25 tokuhirom joined 23:51 zakharyas joined