Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-04-30

| Channels | #pdl index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:15 pdurbin d'oh! I was glad it was logging #gitlove :(
00:16 pdurbin I had asked the founder of #gitlove if it was going to be logged: https://twitter.com/philipdur​bin/status/459559794245120000
00:17 perigrin hopefully it'll be rectified.
00:18 perigrin sungo pointed me at something recently...
00:20 pdurbin sungo?
00:29 perigrin the lead network admin for irc.perl.org
00:31 pdurbin ah
00:32 pdurbin perigrin: pointed you at what?
00:32 perigrin something quite a bit like https://scrollback.io
00:33 pdurbin kinda like http://irclog.perlgeek.de from ilbot2
00:34 perigrin except as a service :)
00:38 pdurbin hmm? :)
00:38 pdurbin oh, a paid service?
00:38 perigrin I want to think it was free but I'm not finding it now
00:39 pdurbin oh, http://irclog.perlgeek.de is free. it logs #crimsonfu at my request
00:42 pdurbin great service. moritz is the best
00:44 perigrin :)
00:44 perigrin he is yes
01:38 perigrin pdurbin: sungo woke up and reminded me it was botbot.me
10:10 pdurbin perigrin: ah yes, that's the second one on my list of irc logging services: http://wiki.greptilian.com/irc/logging
12:48 mohawk joined #pdl
13:59 gtodd joined #pdl
16:22 rindolf joined #pdl
17:18 mohawk joined #pdl
19:27 chm joined #pdl
19:29 chm mohawk: Got your "ears" on?
19:35 mohawk yes i do
19:35 mohawk hello!
19:35 mohawk chm ;)
19:35 chm hi, have you been following my struggles with 64bit fixing?
19:35 chm Any ideas?
19:35 mohawk i have
19:36 mohawk wait, let me see latest
19:36 chm I've found a simple use case but don't have the experience
19:36 chm with C or perl low level debugging to (easily) figure where to go from here.
19:36 chm No debugger, no debugging perl, ...
19:37 mohawk i have a debugging perl built, which takes a while but only needs one time
19:38 mohawk i'm not a perlguts expert, but i've been able to get some stuff done by lots of googling
19:39 chm The lack of a debugger and debugging perl are my main problems.
19:39 mohawk in your example, could you also try $dpdl - $d?
19:39 chm Various reasons but basically my platform (cygwin + win7) is not conducive
19:39 mohawk i don't see that combo in your email
19:39 mohawk i have win7, but i do all my dev now in a linux VM
19:40 chm to getting debugging tools running....
19:41 mohawk additional question: this bad behaviour ($dpdl - 2.0 wrong) - is that the case in the release version?
19:41 chm $dpdl - $d == 0
19:42 chm This issue is only in the longlong-double-fix branch since the
19:42 chm code to address the double vs 64bit int stuff was pretty messy.  I don't
19:42 chm plan to release until, at least, all the tests pass.
19:43 chm If you have a debugging perl, could you try building the longlong-double-fix branch
19:43 chm and see if you can see where the weirdness is coming from?
19:44 mohawk if i did, i wouldn't use a debugger to track it
19:44 mohawk i'm a "printf" kind of guy (or warn in perl)
19:44 chm The debugging perl is the one that you can set flag options for
19:44 chm various levels of diagnostic outputs---not the one that is gdb enabled.
19:44 mohawk so a precise statement of the problem here is that $dpdl - 2.0 is wrong? or is there anything else?
19:45 chm At least that is what I meant  :-)
19:45 mohawk right, i don't have that built i'm afraid
19:45 mohawk i only have -g on mine
19:45 chm Oh.  Yes, $dpdl - 2 is right and $dpdl - 2.0 is wrong (something is truncating the double const to zero I think)
19:48 mohawk i don't know enough about perl insides to know whether that's ever a scalar
19:48 mohawk i'm assuming not, since the actual scalar $d works correctly
19:49 mohawk does the pdl do overloading at all? i'd instrument that?
19:49 chm Since the piddle is involved, I think the 2.0 needs to be promoted to a piddle for the operation.
19:49 chm That is in our code.  I just don't know where and what is being called or why...
19:52 mohawk as you can imagine, i don't really know inside the PDL code at all ;)
19:53 chm This is perl and perl/XS stuff not particular to PDL except that this is the implementation with the bug...
19:53 chm Well, I've got to get to some things here.  Let me know if you try a PDL branch build.
19:54 chm thanks for the thoughts.
19:54 chm o/
19:55 mohawk have you tried doing this:
19:55 mohawk perl -MPDL -de 'print pdl(2) - 2.0, "\n"'
19:55 mohawk and then just "s", then return and follow?
19:57 chm Lots of stepping and n-ing.  No luck so far.  I've probably been looking at the code for too long
19:57 mohawk in that case, take a break? ;)
19:58 chm I guess so---really hoping to get this done before this Sunday's PDL hackathon.
19:58 chm o&o
21:28 mohawk joined #pdl
21:33 mohawk joined #pdl

| Channels | #pdl index | Today | | Search | Google Search | Plain-Text | summary