Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-09-03

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

All times shown according to UTC.

Time Nick Message
15:02 mat_ joined #pdl
15:03 mat_ I'm trying to upgrade PDL on FreeBSD, and I can't seem to get it to build, build log at https://pkg.absolight.fr/data/10amd64-aragor​n/20140903T165016Z/logs/errors/PDL-2.007.log
15:03 mat_ error at the end:
15:03 mat_ Browser.xs:413:9: error: expected expression
15:03 mat_ browse(,
15:04 mat_ maybe I'm missing some module to make it build properly, but I can't figure it out
15:17 mohawk what's the XS code that gens that portion?
15:24 mat_ well, it's IO/Browser/Browser.xs
15:24 mat_ the line is:
15:24 mat_ browse(,
15:24 mat_ __privtrans->__n_size,__pri​vtrans->__m_size,a_datap);
15:25 mat_ I'm not sure exactly how the .xs gets generated though
15:31 mohawk good lord
15:32 mohawk let's take a look at the little fella
15:33 mohawk IO/Browser/browser.pd
15:34 mohawk looks like \$TBSULQFD($PDL_B,$PDL_S,$PDL_U​S,$PDL_L,$PDL_LL,$PDL_F,$PDL_D ) gets evaluated to empty string
15:40 vicash is IO::Browser part of PDL ? i dont think so
15:43 mat_ I have no idea
15:44 mat_ it's in the source :-)
15:45 mohawk vicash, look in the distrib at the file i just pasted name of
15:47 vicash mohawk: by default it should not build
15:47 vicash if you look at perldl.conf which has the line WITH_IO_BROWSER => 0 as the default in the distribution
15:48 vicash is FreeBSD changing that to 1 ?
15:48 mat_ well, yes, it always has
15:49 mat_ I have no idea what it does, but if some part of PDL is broken, it should be removed :-)
15:49 vicash ok. Linux is not changing it to 1 it seems like.
15:49 vicash it may not be broken per se but the testing may not be done enough on freebsd
15:49 vicash because it is set to 0 by default
15:50 vicash mat_ are you building this for a FreeBSD port or installing from CPAN ?
15:51 mat_ vicash, I'm upgrading the FreeBSD port, it's 2.4.11 right now, and I need a newer version for Gimp-Perl
15:51 mohawk how fundamental to PDL is this browser thing?
15:51 mat_ I have no idea
15:52 mohawk mat_, wasn't expecting you to know ;-) clearly i don't either
15:52 mohawk vicash?
15:52 vicash mohawk i have never used it. if it is not required by Gimp-perl then you should skip it. it does not get built by default if you install on CPAN or if you look at the debian/perldl.conf which means a large Linux system just ignores it
15:52 mohawk mat_, does it build/work if you turn off the browser build?
15:52 vicash so maybe FreeBSD should also turn it off and see if shit works fine without it
15:52 mohawk vicash, sounds like yet another part of PDL that should be made a separate module
15:52 vicash then open a ticket with chm and friends and fix the bug.
15:53 mat_ well, we used to patch IO/Browser/Makefile.PL to make it not barf it does "require PDL::Core::Dev;"
15:53 vicash or email perldl and see if there are users who got around it.. there are tons of latent PDL users who dont speak unless they have to
15:54 mat_ I'm going to not build it and see how things goes
15:54 vicash mat_ just use sed and patch perldl.conf WITH_IO_BROWSER to 0 if it is 1
15:54 vicash it should be 0 by default
15:54 vicash and thus IO/Browser/Makefile.PL should never get invokes
15:54 vicash invoked*
15:54 mat_ it is, used to have -e '/WITH_IO_BROWSER/ s|0|1|' -e '/WITH_DEVEL_REPL/ s|1|0|'
15:54 mat_ too
15:55 vicash WITH_DEVEL_REPL should be fine.
15:55 mat_ as I have no idea what PDL is, I'm just trying to go forward :-)
15:55 vicash WITH_DEVEL_REPL 1 should be fine i mean. that works
15:55 vicash unless you dont need it as it makes the pDL drop bigger
15:55 mohawk so sounds like remove the browser mod
15:55 mohawk i would say if freebsd doesn't have PDL on, it probably should
15:56 mohawk it's a powerful piece of software
15:56 mat_ it has PDL, it has 2.4.11 right now
15:56 vicash yes remove the browser mod for now and move on. that browser.pd was written in 1997. maybe it is outdated and no one uses it
15:57 mat_ and Gimp needs a more modern version, so I'm upgrading it
15:57 vicash or maybe you should email perldl and ask how to fix it
15:57 mat_ it seems a bigger beast than perl itself :-)
15:58 mohawk beast is certainly the right word
15:58 mat_ ok, removing PDL::IO::Browser fixed the build
15:58 mat_ (well, not forcing it to build, that is)
15:58 vicash although i am looking at Changes_CVS and Changes and IO::Browser was definitely fixed to build on freebsd
15:59 vicash so maybe it needs to be fixed again. you should definitely open a ticket to avoid breaking applications that may depend on it
15:59 vicash or email the perldl mailing list
16:00 mat_ well, it seems it builds with 5.20 but not 5.16
16:01 vicash yes please email perldl and someone will respond for sure...
16:01 mat_ sure :-)
16:48 mohawk silence implies progress here
17:15 mohawk (i hope)
17:21 chm joined #pdl
17:23 chm mat_, vicash, mohawk: PDL::IO::Browser is an interactive curses-based viewer for 2D pdl data.
17:24 chm It was set by default to not build since the curses detection didn't work except for unix-ish platforms.
17:25 chm The build problem with PDL-2.007 is a bug in the upgrade to 64bit index support.
17:25 chm As far as perl uses for PDL::IO::Browser, AFACT it is not needed for any computation or perl use of PDL
17:26 chm I would like to see a more general PDL visualization tool provided with PDL (OpenGL based or ...?)
17:49 drrho joined #pdl

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