Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-04-19

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

All times shown according to UTC.

Time Nick Message
03:31 krismatrix joined #pdl
04:36 krismatrix I remember asking this question before...but i'd like help with this again. How do I find all the functions/methods available for a particular module like Math::Symbolic?
04:39 preaction perldoc Math::Symbolic
04:42 krismatrix preaction...that only tells me I can use parse_from_string?
04:42 krismatrix I understand that there are more?
04:42 preaction there's a lot more than just the synopsis
04:43 preaction like the SEE ALSO section
04:43 krismatrix isn't the perldoc the same things as in cpan website?
04:44 preaction metacpan and search.cpan.org both display HTML versions of the perldoc, yes
04:44 krismatrix ok..that is easier to read
11:04 mohawk chm, the Alien::Proj4 is required for PDL to build under the next version of EUMM
16:37 krismatrix joined #pdl
16:58 shadowpaste "krismatrix" at 217.168.150.38 pasted "warning uninitialized" (10 lines) at http://paste.scsys.co.uk/473059
16:59 krismatrix I am not sure why it is giving a unitialized value...for this Math::Symbolic script
16:59 krismatrix note...$rhs = 2+3j
17:02 mohawk sivoais, ^
17:02 mohawk sivoais may not reply - he's being lazy and "working" on a "paper" that's important to his "career"
17:03 estrabd joined #pdl
17:04 krismatrix that is fine..my stuff is not urgent anyway...
17:04 krismatrix I need to get of irc...be back in a few mins
17:07 krismatrix joined #pdl
17:07 krismatrix back
17:14 sivoais krismatrix: the issue is that the parser isn't throwing a nice error on the string
17:14 sivoais parse_from_string() is returning undef
17:14 krismatrix how can it be undef?
17:15 sivoais because the syntax of $rhs doesn't match what Math::Symbolic supports
17:15 sivoais if you tried $rhs = "2+3*j", it would work fine
17:15 sivoais Math::Symbolic doesn't support implicit multiplication like that
17:16 krismatrix like what exactly...you mean if $rhs = 2+3j versue $rhs = "2+3j"?
17:18 sivoais my $rhs = "2+3j" # doesn't work
17:18 sivoais my $rhs = "2+3j" # doesn't work because Math::Symbolic can't parse it
17:19 sivoais my $rhs = 2+3j # no quotes: doesn't work because that isn't valid Perl syntax
17:19 sivoais my $rhs = "2+3*j" # works because Math::Symbolic can parse it
17:19 krismatrix ah...ok...i got it to work
17:20 chm joined #pdl
17:20 krismatrix It seems that z = 2 + 3*j work...it doesn't have to be in quotes
17:21 sivoais oh, I wasn't sure if you were trying to Perl syntax to assign that to the variable
17:21 sivoais :-P
17:22 chm mohawk: Why is Alien::Proj4 is required for PDL to build under the next version of EUMM and what does EUMM releases have to do with the PDL build requirements?  Shouldn't they be orthogonal?
17:25 krismatrix Well...I am trying to do a few things....I want user to enter z = 2 + 3*j
17:25 krismatrix Then I put z in $lhs for left hand side eqn
17:25 krismatrix Then I put 2 + 3*j in $rhs for right hand side equation
17:25 krismatrix And of course parse the $rhs
17:26 krismatrix I think I may need to use OO for this section...because I need my program to maintain variable values? What do you think?
17:28 sivoais krismatrix: I would use a data structure first (symbol table) then move to making a OO wrapper around it
17:29 krismatrix hmm...i don't know what a data structure is....let me see what I can find on that
17:37 sivoais krismatrix: hash tables, linked lists, etc. Those are all data structures
17:37 sivoais in this case, I'd just use a Perl's built-in hash
17:41 krismatrix I know about hash....is hash table something different? Any I recall learning linkedlist in college but don't remember them anymore.
17:44 sivoais hash tables and Perl hashes are the same thing. And you honestly don't need linked lists unless you're coding at a low level of abstraction :-P
17:44 krismatrix oh..ok...that makes sense...
17:45 krismatrix I know how to use Hashes just fine...Wrapping my head around OO is another matter...:)
17:46 sivoais hehe, later you'll want to learn about Moose and Moo. They make OO easy to use in Perl. But writing good OO code is a different skill :-)
17:50 mohawk chm, it's because part of PDL (P::T::Proj4) depends on another part of it (P::GIS::Proj) already being built, which used to work but thanks to a GNU make bug and a change in EUMM doesn't reliably work
17:51 mohawk the most immediate problem you're having is due to (i think) EUMM having difficulty with quoted command line flags (which are necessary to work on "perl in space") - it just needs a fix to EUMM
17:52 mohawk krismatrix, OO is often more trouble than it's worth
17:52 mohawk unless you're trying to create something that others can reuse
17:52 mohawk even then it's easy to get wrong
17:53 krismatrix Well..as i have told you guys before...i am trying to write a tool that will do Elec Engr computations...in order to do that...i need to do Math comps.
17:53 krismatrix So I need something that works like Octave on my program
17:53 krismatrix i.e. similar syntax and functionality
17:54 mohawk krismatrix, you're going to need to write a mathematical formula parser
17:54 mohawk it will need to make a parse table, like sivoais said
17:54 krismatrix isn't that what Math::Symbolic is doing?
17:54 krismatrix i.e. parsing the user input?
17:54 mohawk it's almost certainly parsing
17:55 mohawk but you don't like its syntax, i gather
17:55 krismatrix No...i am fine with that ...at least so far
17:55 krismatrix I don't math Symbolic well enough to really have a big opinion..i am learning as I go
17:55 krismatrix don't know Math Symbolic
17:56 krismatrix Basically..i am starting by being able to do something like this
17:56 krismatrix User input> z = 2 +3*j +  4*j
17:56 krismatrix prints  z = 2 + 7*j
17:57 krismatrix need to get from user input to output
17:58 chm mohawk:  That complicates things for a quick fix since the Alien::Proj doesn't support the ability to use perldl.conf settings to find and use an installed PROJ4 setup.
17:59 mohawk chm, it exactly does support that
17:59 mohawk that is the entire point of it
18:00 chm mohawk:  It exactly doesn't support that since I have a cygwin build that now fails but works for 2.007 and is broken now.  I thought I sent an email to you with my findings in the hope you might figure out what is happening.
18:00 mohawk the problem here is, i believe, the failure of EUMM's lib-hunting code to deal correctly with "" args
18:00 mohawk i received that email
18:01 mohawk that is the basis on which i am saying that is my belief re the problem
18:01 chm yes, the quoting seemed to be the problem.  As an EUMM devel, isn't there a "quick fix" we could use for PDL-2.008 to avoid the regression and give us a chance to really fix things up for 2.009?
18:02 mohawk i am reading the code now to find out
18:03 chm mohawk:  And believe me I am familiar with " challenges and libraries for EUMM.  For example, if perl was linked with a library then they are stripped from the link line even though the XS does not have the link.  :-)
18:04 mohawk this could even be cygwin-specific
18:04 mohawk chm, is there a way for you to give me a remote login on your cygwin setup?
18:09 chm mohawk:  I'm not set up for that at the moment.
18:10 chm The quickest might be to install libproj in a location on your cygwin to see if you can reproduce.
18:10 mohawk i don't have cygwin
18:10 chm You might even be able to track things down just with setting perldl.conf and running perl -d Makefile.PL with $DB::single=1 in the various files.
18:10 chm Oh, try the second approach then...
18:11 chm I'm actually on a different cygwin setup than previously.  I'll test this one and let you know.
18:12 mohawk i believe this is cygwin-specific since it works for sisyphus on his win32
18:12 chm It could also be EUMM version related...
18:13 chm Did the EUMM required change?  Let me double check here.
18:17 mohawk ExtUtils::Liblist (and ::Kid) are a godawful mess
18:17 mohawk they have not yet received the cleansing fire of my justice
18:19 mohawk i have an idea for a simple quick fix
18:19 mohawk only put "" around things with spaces
18:24 opkick [pdl] mohawk2 created alienquickfix (+1 new commit): http://git.io/vvhif
18:24 opkick pdl/alienquickfix 668bd37 Ed J: Work around EU::Liblist::Kid not working with "-L" flags on Cygwin
18:24 opkick [pdl] mohawk2 opened pull request #100: Work around EU::Liblist::Kid not working with "-L" flags on Cygwin (master...alienquickfix) http://git.io/vvhiJ
18:24 mohawk chm, please could you give that branch a go?
18:24 mohawk if it works for your situation and passes travis and my win32, it should probably be merged and released as another RC
18:26 mohawk (i'm running the win32 test now)
18:33 mohawk chm, please try the above
18:34 mohawk good news: it's coming up green lights on travis on bleading edge EUMM and installed, so that's good news
18:35 chm mohawk, I was able to reproduce the fail from master.  I'll test your fix next.  May be an hour or so..
18:36 mohawk if you kick it off now, it can run in background? :-)
18:36 chm If I were here to kick it off.  Later. o/
18:36 opkick [pdl] mohawk2 comment on issue #100: Passes on Win32. http://git.io/vvh11
19:14 opkick [pdl] coveralls comment on issue #100:
20:40 chm joined #pdl
20:41 chm mohawk:  All tests PASS *and* PROJ is detected and built with.
20:46 mohawk great
20:47 mohawk i'll merge, then you release RC2?
20:47 chm mohawk: OK
20:48 opkick [pdl] mohawk2 closed pull request #100: Work around EU::Liblist::Kid not working with "-L" flags on Cygwin (master...alienquickfix) http://git.io/vvhiJ
20:48 opkick [pdl] mohawk2 merged alienquickfix into master: http://git.io/vvjYt
20:48 mohawk done
20:51 chm Thanks, I'll roll and announce another RC to CPAN
20:53 mohawk chm++
21:25 opkick [pdl] wchristian tagged v2.007_15 at d08da06: http://git.io/vvjc7
21:30 opkick [pdl] wchristian pushed 1 new commit to master: http://git.io/vvjWv
21:30 opkick pdl/master 85e8e2a Chris Marshall: Update VERSION for final release preparations
21:31 mohawk chm, aren't we on RC2?
21:31 mohawk oh, it's post _15, gotcha
21:33 chm mohawk: On the list: standardize on release/tags/naming
21:33 mohawk no, it's fine
21:34 mohawk was only questioning the number part, not the concept ;-)
21:35 chm Thanks the quick help today.  With luck, there will be no egregious FAILs from cpan testers and we can get to the final wordsmithing for 2.008.
21:36 travis-ci PDLPorters/pdl#190 (master - 668bd37 : Ed J): The build passed.
21:36 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/ecb2008a3d2f...668bd377c0eb
21:36 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/59157871
21:36 mohawk i'm still concerned about the 3 new fails on mac
21:37 mohawk but since the reporter hasn't added any info on whether the same test would fail on 2.007, i say let us bravely plough on
21:37 mohawk the tests are now far better than they have ever been, largely thanks to sivoais
21:37 mohawk when we've got 2.008 out, we can split the iceberg!
21:38 chm As you are aware, more tests mean more ways to fail!  :-)  Of course it also means things can change more with less risk.
21:38 mohawk it only appears to be failure - it's really letting us know earlier rather than once in production ;-)
21:38 mohawk but smaller parts means more agility
21:38 chm If there are some outstanding FAIL for mac platforms, not on 2.007, we should take a look.
21:38 mohawk well we don't know if it's not on 2.007 or not
21:39 chm Are these travis fails or cpan testers fails?
21:39 chm Either way, we should be able to check.
21:40 mohawk it was in an email
21:40 mohawk let me check back...
21:43 mohawk hmm
21:43 mohawk sivoais, do you remember - the thing with t/clump.t, and ones.t?
21:43 chm mohawk: I just noticed a bunch of UNKNOWN ("fails") from missing File::Path.  Is that new?  Does it need to be in prereqs?
21:43 mohawk maybe it was a FAIL report
21:43 mohawk chm, example URL?
21:44 mohawk File::Path has been in core since 5.001
21:44 mohawk something is badly wrong if it's missing
21:45 chm Just reporting what I'm seeing.  It is a convenience to assume core modules.  Technically, it would be correct to put dependencies on all modules "use-ed"
21:46 mohawk if an installation is that badly broken, it's not something we can be expected to deal with
21:48 chm Working around problems of other installations is part of making releases work.
21:48 chm I would prefer everything simple and clear but the result is FAIL and UNKNOWN results from cpan testers that suggest *not* to use PDL;
21:49 chm Besides, adding it explicitly is not wrong, just shouldn't be needed.
21:49 chm However, I don't think there is any requirement that perl releases only include the official core modules.
21:49 mohawk adding it won't help
21:50 chm And why not?
21:50 mohawk the NAs i looked at are on installs with Filter::Simple < 0.86
21:50 mohawk it runs the build anyway then gives this "NA" result
21:50 chm I'm talking about the UNKNOWN with File::Path
21:50 mohawk right
21:51 mohawk so it'll given an NA
21:51 chm I'm going to look into the Filter::Simple problem as well.  Just not this PM.  Got to get to dinner.
21:51 chm Thanks again, mohawk.  o>
21:51 mohawk later amigo
21:51 chm o/
21:55 mohawk now "cpanm --dev --test-only PDL" works, thanks to some deleting old stuff by CSOE and KGB
22:06 mohawk this might be useful: the big number modulus only seem to fail on i686
22:06 mohawk http://www.cpantesters.org/distro/P/PDL.​html?grade=2&amp;perlmat=2&amp;patches=2​&amp;oncpan=1&amp;distmat=1&amp;perlver=​ALL&amp;osname=ALL&amp;version=2.007_14
22:07 mohawk (and i386)
22:20 travis-ci PDLPorters/pdl#191 (master - c171158 : Chris Marshall): The build has errored.
22:20 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/668bd377c0eb...c171158be4fd
22:20 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/59161587
22:21 mohawk don't panic!
22:21 mohawk that was because github timed out, i restarted
22:35 travis-ci PDLPorters/pdl#191 (master - c171158 : Chris Marshall): The build passed.
22:35 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/668bd377c0eb...c171158be4fd
22:35 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/59161587
22:42 mohawk there ya go
23:08 travis-ci PDLPorters/pdl#192 (master - 85e8e2a : Chris Marshall): The build passed.
23:08 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/c171158be4fd...85e8e2a74ed5
23:08 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/59161919
23:09 mohawk chm, SF bug 377 is dead and buried
23:39 sivoais mohawk: the only place I saw those fails on Mac was jberger's paste in here a few days back
23:39 sivoais I hadn't looked at CPAN Testers yet
23:55 sivoais oh wow... hail in Houston

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