Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-03-14

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

All times shown according to UTC.

Time Nick Message
02:16 opkick [Alien-HDF4] mohawk2 pushed 1 new commit to master: http://git.io/pddz
02:16 opkick Alien-HDF4/master bce22a2 Ed J: v0.02 - BUILD_REQUIRES -> CONFIGURE_REQUIRES
02:17 travis-ci PDLPorters/Alien-HDF4#9 (master - bce22a2 : Ed J): The build passed.
02:17 travis-ci Change view : https://github.com/PDLPorters/Alien-HD​F4/compare/eba7c566b990...bce22a2c083a
02:17 travis-ci Build details : http://travis-ci.org/PDLPorte​rs/Alien-HDF4/builds/54329997
02:21 opkick [Alien-HDF4] mohawk2 pushed 2 new commits to master: http://git.io/pdFY
02:21 opkick Alien-HDF4/master a63b1ca Ed J: Add .gitignore
02:21 opkick Alien-HDF4/master 7c17fa3 Ed J: v0.03 - copy BUILD_REQUIRES to CONFIGURE_REQUIRES
02:23 travis-ci PDLPorters/Alien-HDF4#10 (master - 7c17fa3 : Ed J): The build passed.
02:23 travis-ci Change view : https://github.com/PDLPorters/Alien-HD​F4/compare/bce22a2c083a...7c17fa398779
02:23 travis-ci Build details : http://travis-ci.org/PDLPorte​rs/Alien-HDF4/builds/54330225
02:31 opkick [pdl] mohawk2 force-pushed alienhdf4 from 1ee4bab to e5e5763: http://git.io/x2TT
02:31 opkick pdl/alienhdf4 e5e5763 Ed J: Use Alien::HDF4 in PDL::IO::HDF build
02:45 opkick [pdl] mohawk2 force-pushed alienhdf4 from e5e5763 to 9ba08a2: http://git.io/x2TT
02:45 opkick pdl/alienhdf4 9ba08a2 Ed J: Use Alien::HDF4 in PDL::IO::HDF build
03:13 mohawk sivoais, got a bit of a travis problem, see http://paste.scsys.co.uk/469402
03:16 sivoais mohawk: looking now
03:16 mohawk shweet
03:17 opkick [pdl] coveralls comment on issue #21:
03:21 sivoais weird and it's using x86_64-linux-thread-multi in @INC, not x86_64-linux
03:21 mohawk but look at the bit before
03:22 mohawk it's 5.20-extras
03:22 mohawk not eg 5.10
03:22 mohawk so the @INC is getting a wrong value
03:22 sivoais ah, how did the -extras even get in there
03:22 mohawk worse
03:22 mohawk 5.***20***-extras
03:22 mohawk not 5.10
03:22 mohawk yes?
03:23 sivoais but the install said it installed to 5.20
03:23 mohawk it did
03:23 sivoais 5.10 isn't even in the paste
03:23 mohawk that's why its search in 5.20 is doomed
03:23 mohawk i know
03:23 sivoais wwwhhhaaat!
03:23 sivoais so, this paste was from the 5.10  build?
03:23 mohawk if you do as i did, and look at another build log, you will see
03:24 mohawk i'll draw your attention to the part of that paste where i said "across all perl versions"
03:24 mohawk that meant builds
03:24 mohawk truthfully i only looked at 5.10 as well as 5.20
03:24 mohawk but i will bet you a shiny silver dollar it's true across all
03:24 sivoais where is the link to the log?
03:25 mohawk https://travis-ci.org/PDLP​orters/pdl/builds/54331533
03:28 mohawk i'll ask haarg on #toolchain, he's a travis guy
03:29 sivoais yeah, he wrote the helpers we're using
03:29 sivoais plus, there's now a debug environment variable you can enable to debug the helpers
03:30 sivoais mohawk: you should also add Alien::HDF4 to the 00-report-prereqs.t
03:32 sivoais mohawk: I'm thinking the system Perl is being picked up over the travis-perl-helper Perl
03:32 mohawk sivoais, if it were, why would it have perlbrew in @INC at all?
03:33 mohawk how do you have in mind to add A::H to the prereqs.t? that's purely using the prereqs, and i haven't added it as a prereq
03:34 mohawk that was deliberate, to match current situation - works if there, ok if not
03:34 mohawk using the prereqs = META.*
03:35 sivoais ah, I see
03:37 mohawk this is one reason why we desperately need to break out eg PDL::IO::HDF into its own distro
03:37 mst joined #pdl
03:37 mohawk PDL becomes the bundle
03:37 shadowpaste joined #pdl
03:37 mohawk PDL::Core becomes the current "Basic" dir
03:37 mst mohawk: try now
03:37 mohawk ok
03:37 mst it needs the bot in the channel
03:38 shadowpaste "mohawk" at 217.168.150.38 pasted "test" (1 line) at http://paste.scsys.co.uk/469403
03:38 mst tada
03:38 mohawk booyakasha
03:38 mst all you have to do is '/invite shadowpaste'
03:38 mohawk thanks!
03:38 mohawk noted
03:38 mst I try and keep it configured to remember but that's manual
03:38 mohawk it's fine
03:38 mst and I usually do that after it's been useful a few times
03:39 mohawk now i know, it's fixable
03:39 mst so basically remind me after the next time you /invite
03:39 mst and I'll make it permanent
03:39 mohawk i could remind you now ;-)
03:39 mohawk it's been in the /topic for a long time
03:39 mohawk we do use it from time to time
03:39 mohawk just first time i noticed it not cooperating
03:40 mohawk mst++ # enabling stuff
03:40 sivoais :-D
03:40 mohawk sivoais, back to fixing travis! no cookie for you (yet)
03:41 sivoais awww :-(
03:41 * mohawk goes back to figuring why MatrixOps, and only that, doesn't build its .so under "make core"
03:42 mohawk you have to EARN the cookie
03:42 mst ok, added to permanent config
03:42 mst left #pdl
03:42 mohawk mst, nice one
03:44 sivoais interesting, mohawk, it seems it was always running build-dist using 5.20-extras
03:44 sivoais see this old job <https://travis-ci.org/PDLPorters/pdl/jobs/53246491> line 2574
03:50 sivoais mohawk: this is the reason: <https://github.com/travis-perl/helpers/comm​it/b4347eef9af53731b3da5506ae3fec9345ef6dde>
03:51 mohawk sivoais, you're a genius
03:51 mohawk how are we going to sort this?
03:53 sivoais hmm, let me see if I can make a patch to travis-perl-helpers
03:53 sivoais build-dist --build-with-current-perl ?
03:53 sivoais just as a temporary name
03:54 sivoais I think the reason for using the latest Perl is because Dist::Zilla plugins often are made for the latest Perl even if the dist being made has a lower min. Perl version
03:54 mohawk sivoais, what if you captured the PERL5LIB before that, and reset it after?
03:54 mohawk may i suggest opening an issue on travis-helpers?
03:55 mohawk first is workaround, second is raising awareness towards a solution
03:55 sivoais yes!
03:55 sivoais doing so now
03:57 sivoais or we could just use perl Makefile.PL directly ;-)
04:02 sivoais actually, the best way would be to switch back to build Perl right before calling Makefile.PL
04:02 sivoais since that would more accurately represent the Dist::Zilla -> CPAN -> user install process
04:04 mohawk sivoais, i hear you
04:04 mohawk i'm not at all concerned with simulating DZ since we don't use it
04:04 mohawk and yes, i see no problem with just using "perl M.PL"
04:04 sivoais yeah, but for the sake of making build-dist correct :-)
04:05 mohawk let that be someone else's problem if it works correctly for us
04:05 mohawk especially since we raised it
04:05 sivoais Let me make that change on a branch, test, merge, and then you rebase?
04:05 sivoais or I can go ahead and rebase so there's no deadlock :-D
04:06 mohawk use your judgement, but i think a branch just to check it
04:06 mohawk branches are good
04:06 mohawk and also to offer the review process
04:09 sivoais added issue at <https://github.com/travis-perl/helpers/issues/17>
04:09 sivoais and now making that branch
04:11 mohawk nice
04:18 opkick [pdl] zmughal created travis-use-eumm-directly (+1 new commit): http://git.io/pFk1
04:18 opkick pdl/travis-use-eumm-directly 78c13ab Zakariyya Mughal: Travis-CI: use perl Makefile.PL directly...
04:18 opkick [pdl] zmughal opened pull request #68: Travis-CI: use perl Makefile.PL directly (master...travis-use-eumm-directly) http://git.io/pFkS
04:18 sivoais let's see how that builds :-)
04:19 sivoais the helpers currently use a sad trick to capture missing deps when using a different Perl :-(
04:20 sivoais <https://github.com/travis-perl/help​ers/blob/cd10be18a0482f66f030d67e0c​eae1d9ea98c7fe/bin/build-dist#L22>
04:20 sivoais > $(perl -lne'if (/\bCan'\''t locate (\S+)\.pm in \@INC/)
04:31 mohawk oh dear
04:31 mohawk i say let the META.* be it
04:31 mohawk and if it fails the authors can add to the *_REQUIRES
04:32 sivoais yeah, that's the best way
04:32 mohawk sivoais, i am happy for us to merge anything travis-orientated if it works as intended
04:32 mohawk ie, if it passes
04:33 sivoais yes, that's what I'm waiting for
04:33 mohawk please feel free to use alien* as a testbed
04:33 sivoais wait, did it fail...!
04:33 mohawk yes it did
04:33 sivoais hmpph
04:33 * sivoais looks
04:33 mohawk in other words, i am suggesting to put an extra commit on alien*
04:34 mohawk and keep amending / pushing that until it works, then cherry-pick it
04:34 mohawk even if i've gone to bed :-)
04:34 sivoais haha, ok
04:35 mohawk although it might be the very wisest to cherry-pick to another branch, check that passes at least one travis build before merging
04:36 sivoais ok, sounds good :-D
04:37 mohawk shweet
04:44 sivoais mohawk: I don't think Devel::CheckLib should be in that Makefile.PL...
04:44 sivoais it's used in Basic/Core/Makefile.PL
04:49 sivoais *sigh* that's a thing for another day, I suppose
04:52 mohawk sivoais, i don't understand - in which M.PL?
04:53 opkick [pdl] zmughal opened issue #69: Devel::CheckLib is a configure dependency but still used directly in Basic/Core/Makefile.PL http://git.io/pFqW
04:53 sivoais toplevel
04:53 sivoais <https://github.com/PDLPorters/pdl/issues/69>
04:54 sivoais all calls to check_lib need to be put in a CONFIGURE coderef, correct?
04:54 sivoais because it might not be installed yet
04:54 mohawk i wouldn't say so
04:54 mohawk i say do CONFIGURE_REQUIRES
04:54 mohawk that makes a META.*
04:54 mohawk and if the installer (cpanm or a human) doesn't respect that, any failure is on them
04:55 mohawk i'm faintly familiar with CONFIGURE, having fought it in EUMM
04:55 mohawk but i've never really seen a serious use for it
04:55 mohawk i don't even see how it would help here?
04:55 sivoais ok, so when does MY::processPL run?
04:56 mohawk whenever processPL would be
04:56 mohawk it's a method override
04:56 mohawk please tell me you're not thinking about getting fancy
04:56 mohawk in fact, what are you trying to achieve?
04:57 sivoais I run the following
04:57 sivoais pm-uninstall Devel::CheckLib; cpanm --installdeps --verbose .
04:57 sivoais > Can't locate Devel/CheckLib.pm in @INC
04:57 sivoais I get that output
05:00 sivoais mohawk: tell me I'm doing something wrong .... haha
05:01 sivoais and removing the Devel::CheckLib from the top-level file just offloads it to Basic/Core/Makefile.PL
05:02 sivoais the only reason it was working before was because Devel::CheckLib was in inc/
05:03 mohawk hold on
05:03 mohawk which kind of REQUIRES are we specifying it in?
05:03 sivoais CONFIGURE_REQUIRES
05:03 sivoais as it should be
05:03 mohawk correct
05:04 mohawk there's a cpanm --installdeps somewhere?
05:04 sivoais but now that I dropped it out of inc/, it needs to be installed by the cpan client before any other Makefile.PL runs
05:04 mohawk ok
05:04 mohawk build-dist probably makes *META.*
05:05 mohawk bit of a bootstrap problem
05:05 sivoais yes, it captures all missing modules and installs them later... being too smaht
05:05 mohawk this is only relevant when we run from in the repo
05:06 mohawk any other environment will have the META.*
05:06 sivoais ahhh, I see
05:06 sivoais make dist DTRT
05:06 mohawk author-side vs install-side, as i call it
05:07 mohawk there's no reason to use D::CL in top-level one, since it doesn't use it
05:07 mohawk use = "use"
05:07 mohawk but obviously it still needs specifying in *_REQ*
05:08 mohawk and also it should be "use"d in each place check_lib* is called
05:08 mohawk $ gg check_lib
05:08 mohawk Basic/Core/Makefile.PL:       if (check_lib(LIB=>'-lpthread',header=>'pthread.h')) {
05:08 mohawk IO/Browser/Makefile.PL:    if (check_lib(header=>$incl)) {
05:08 mohawk IO/Browser/Makefile.PL:   if (check_lib(lib=>$libr)) {
05:08 mohawk Lib/GIS/Proj/Makefile.PL:         check_lib(
05:08 mohawk Lib/Transform/Proj4/Makefile.PL:    if (check_lib(
05:08 sivoais the GitHub user named INC must really not like Perl programmers...
05:08 mohawk ha ha ha
05:08 mohawk probably a perl programmer who made a bad life choice
05:09 mohawk ok, stepping up for a second - the thing we're looking at right now is important for future gh dev
05:09 mohawk it is not at all important for 2.008
05:09 mohawk what could we be doing that advances 2.008?
05:09 sivoais pinging chm! :-D
05:11 opkick [pdl] zmughal closed issue #69: Devel::CheckLib is a configure dependency but still used directly in Basic/Core/Makefile.PL http://git.io/pFqW
05:13 mohawk wo, hang on
05:13 mohawk that's what configure deps are for
05:14 sivoais O_o
05:14 mohawk EUMM is another
05:14 * sivoais nods
05:14 mohawk on author-side, it's up to us to either manually install known deps, or spot "couldn't load" messages
05:14 mohawk the travis helper is correct here and we/you are not
05:15 sivoais ok, so adding Devel::CheckLib before is the fix or no?
05:15 mohawk sadly, the only way to make travis work sanely is to be author-side (rather than impose the reqmt of having to check in install-side stuff)
05:15 sivoais cpanm Devel::CheckLib
05:15 mohawk yes
05:15 mohawk that's the way
05:15 sivoais ok, that's what I'm about to push! :-)
05:15 mohawk it's simulating what cpanm would do with the META.*
05:16 mohawk nice work ;-)
05:17 sivoais ah, so we should close <https://github.com/travis-perl/helpers/issues/17>...
05:18 sivoais or at least start adding a "My build is doing this weird thing" FAQ to the docs
05:19 mohawk no, i think it's a useful conversation to have
05:19 mohawk there may be a better way
05:19 * sivoais nods
05:19 mohawk CONFIGURE_REQUIRES means "my Makefile.PL will blow up without this"
05:19 mohawk so expecting the M.PL to run and tell us what it needs is a bit misguided
05:20 mohawk it's a genuine chicken/egg situation
05:20 sivoais so the proper way is two passes ;-)
05:20 sivoais eep
05:20 mohawk at least two
05:20 mohawk hence the t-h regex to spot missing modules
05:26 opkick [pdl] zmughal force-pushed travis-use-eumm-directly from 78c13ab to 22ecdce: http://git.io/pFOP
05:26 opkick pdl/travis-use-eumm-directly 22ecdce Zakariyya Mughal: Travis-CI: use perl Makefile.PL directly...
05:28 sivoais gave it a better commit message to say why we need to run the CONFIGURE_REQUIRES before
05:29 mohawk sivoais++
05:30 mohawk export BUILD_DIR="$(pwd)/build_dir"; perl Makefile.PL DISTVNAME=$BUILD_DIR
05:30 mohawk Makefile.PL: OpenGL-0.6702 not found, setting $PDL::Config{USE_POGL} => 0
05:30 mohawk Types.pm.PL: using typedef long PDL_Indx
05:30 mohawk Extracting Types.pm
05:30 mohawk making PDL_B...
05:30 mohawk making PDL_S...
05:30 mohawk making PDL_US...
05:30 mohawk making PDL_L...
05:30 mohawk making PDL_IND...
05:30 mohawk making PDL_LL...
05:30 mohawk that's interesting
05:30 mohawk i didn't copy that
05:31 mohawk The command "cd $BUILD_DIR" failed and exited with 1 during .
05:31 mohawk that, i did
05:31 mohawk it's failing: https://travis-ci.org/PDLPorters/pdl/jobs/54338168
05:32 sivoais so it writes to the BUILD_DIR and the it's gone?
05:32 * sivoais investigates
05:38 opkick [pdl] zmughal force-pushed travis-use-eumm-directly from 22ecdce to 6a4f9b0: http://git.io/pFOP
05:38 opkick pdl/travis-use-eumm-directly 6a4f9b0 Zakariyya Mughal: Travis-CI: use perl Makefile.PL directly...
05:39 sivoais get on it, Frere Travis!
05:43 opkick [pdl] zmughal force-pushed travis-use-eumm-directly from 6a4f9b0 to 4aebf5d: http://git.io/pFOP
05:43 opkick pdl/travis-use-eumm-directly 4aebf5d Zakariyya Mughal: Travis-CI: use perl Makefile.PL directly...
05:48 opkick [pdl] zmughal force-pushed travis-use-eumm-directly from 4aebf5d to ab7b3a8: http://git.io/pFOP
05:48 opkick pdl/travis-use-eumm-directly ab7b3a8 Zakariyya Mughal: Travis-CI: use perl Makefile.PL directly...
06:15 opkick [pdl] coveralls comment on issue #68:
06:49 opkick [pdl] wchristian created alienhdf4-use-eumm (+1 new commit): http://git.io/pFlU
06:49 opkick pdl/alienhdf4-use-eumm 2320795 Zakariyya Mughal: Merge branch 'travis-use-eumm-directly' into alienhdf4-use-eumm...
06:50 opkick [pdl] zmughal opened pull request #70: Alien::HDF4 and use EU::MM directly (master...alienhdf4-use-eumm) http://git.io/pFlY
07:17 opkick [pdl] coveralls comment on issue #70:
07:50 opkick [Alien-HDF4] zmughal deleted travis-ci at eba7c56: http://git.io/pFzZ
13:07 mohawk sivoais, that's a lot of branches! where are we at?
16:18 sivoais mohawk: I believe just using EUMM directly works, but I will try the SPLIT_BUILD=0 method
16:56 opkick [pdl] zmughal created travis-disable-split-build (+1 new commit): http://git.io/pAe1
16:56 opkick pdl/travis-disable-split-build 263d8ae Zakariyya Mughal: Travis-CI: set SPLIT_BUILD to false to use the run's version of Perl for configure-time...
16:57 opkick [pdl] zmughal opened pull request #71: Travis-CI: set SPLIT_BUILD to false to use the run's version of Perl for configure-time (master...travis-disable-split-build) http://git.io/pAvl
16:58 opkick [pdl] zmughal comment on issue #71: If this passes, it obsoletes #68. http://git.io/pAvM
17:09 opkick [pdl] zmughal created alienhdf4-disable-split-build (+1 new commit): http://git.io/pAUD
17:09 opkick pdl/alienhdf4-disable-split-build cc1199c Zakariyya Mughal: Merge branch 'travis-disable-split-build' into alienhdf4-disable-split-build...
17:10 opkick [pdl] zmughal opened pull request #72: Alien::HDF4 and disable SPLIT_BUILD (master...alienhdf4-disable-split-build) http://git.io/pATs
17:10 opkick [pdl] zmughal comment on issue #72: This obsoletes #70 by using #71. http://git.io/pATW
17:23 opkick [pdl] coveralls comment on issue #71:
17:28 sivoais mohawk: looking good so far
17:37 opkick [pdl] zmughal closed pull request #68: Travis-CI: use perl Makefile.PL directly (master...travis-use-eumm-directly) http://git.io/pFkS
17:38 opkick [pdl] zmughal deleted travis-use-eumm-directly at ab7b3a8: http://git.io/pAsR
17:39 opkick [pdl] zmughal comment on issue #70: Closed by #72. http://git.io/pAs5
17:40 opkick [pdl] zmughal deleted alienhdf4-use-eumm at 2320795: http://git.io/pAGk
17:47 mohawk sivoais, where are we?
17:47 sivoais mohawk: waiting on the last builds of Travis, but I say that travis-disable-split-build is good to go
17:47 sivoais i.e., merge into master
17:48 sivoais and then alienhdf4 can rebase off that
17:49 sivoais I'm curious why we're not getting any more build success messages in here
17:50 sivoais \o/ ok, looks like alienhdf4-disable-split-build is all good
17:50 opkick [pdl] coveralls comment on issue #72:
17:52 opkick [pdl] zmughal merged travis-disable-split-build into master: http://git.io/pAC3
17:53 opkick [pdl] zmughal comment on issue #72: Test done. Can now delete this. http://git.io/pAC4
17:53 mohawk sivoais++
17:53 mohawk rebasing now
17:53 opkick [pdl] zmughal deleted alienhdf4-disable-split-build at cc1199c: http://git.io/pACg
17:55 mohawk sivoais, if you've merged, delete travis-disable-split-build?
17:56 sivoais done!
17:56 opkick [pdl] zmughal deleted travis-disable-split-build at 263d8ae: http://git.io/pAl8
18:00 opkick [pdl] mohawk2 force-pushed alienhdf4 from 9ba08a2 to b4220e8: http://git.io/x2TT
18:00 opkick pdl/alienhdf4 b4220e8 Ed J: Use Alien::HDF4 in PDL::IO::HDF build
18:03 mohawk sivoais, now let's see how that plays :-)
18:10 * sivoais & # needs to go work for a couple hours :-)
18:18 mohawk sivoais, enjoy! chat later
18:22 travis-ci PDLPorters/pdl#108 (master - 263d8ae : Zakariyya Mughal): The build passed.
18:22 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/d751ac7eb426...263d8aeb12ea
18:22 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/54380823
18:45 mohawk sivoais, bad news: it's still shows 5.20-extras and failing to find Alien::HDF4: https://travis-ci.org/PDLPorters/pdl/jobs/54381465
18:46 opkick [pdl] coveralls comment on issue #21:
19:01 opkick [pdl] mohawk2 pushed 1 new commit to alienhdf4: http://git.io/pA91
19:01 opkick pdl/alienhdf4 8475732 Ed J: Move SPLIT_BUILD setting to before init, where it is used
19:04 mohawk sivoais, now THAT works beautifully
19:04 mohawk ;-)
19:04 mohawk https://travis-ci.org/PDLPorters/pdl/jobs/54386288 shows "Building PDL::IO::HDF. Turn off WITH_HDF if there are any problems"
19:06 mohawk (reason is, "init" is where the SPLIT_BUILD is used, to set the value of MODERN_PERL - setting it just for build-dist is too late)
19:08 sivoais mohawk++
19:10 sivoais mohawk: I'm going to cherry-pick that onto master
19:10 sivoais sound good?
19:10 sivoais or just wait until alienhdf4 is merged?
19:17 mohawk no, cherry pick now
19:17 mohawk otherwise we have ineffective changes there :-)
19:17 mohawk sivoais, ^
19:25 mohawk this also explains why the ParseXS changes weren't working
19:25 mohawk god knows how the --force changed anything
19:27 opkick [pdl] coveralls comment on issue #21:
19:27 sivoais oh my... yes
19:28 mohawk anyway, cherry-pick away :-)
19:28 mohawk then i can rebase and re-push
19:29 mohawk (this functions as me making a change, you reviewing then merging, so it works for me)
19:31 opkick [pdl] zmughal pushed 1 new commit to master: http://git.io/pAxX
19:31 opkick pdl/master 8647162 Ed J: Move SPLIT_BUILD setting to before init, where it is used
19:32 sivoais heh, there's no Inline::Ada yet? There should be!
19:35 mohawk let's finish PDL 2.008 first ;-)
19:35 mohawk so, where are we on that?
19:35 mohawk sivoais++ # shepherding this crazy travis thing
19:35 opkick [pdl] mohawk2 force-pushed alienhdf4 from 8475732 to 0000530: http://git.io/x2TT
19:35 opkick pdl/alienhdf4 0000530 Ed J: Use Alien::HDF4 in PDL::IO::HDF build
19:36 sivoais let me e-mail the mailing list again
19:36 mohawk nice
19:36 mohawk one thing to bear in mind
19:37 mohawk lldfp1 changed the core version
19:37 mohawk that means people will have to recompile
19:37 sivoais I believe that is the issue with all of the 2.007 DEV releases
19:38 mohawk can we confirm with which dev release the core version changed from 2.007, then?
19:38 sivoais yes, I believe a I saw a release note with that info
19:39 mohawk this is where people look in the Changes file
19:39 mohawk we do need to switch that
19:39 mohawk i think a way forward on that will be to use github's "compare" functionality to read through the changes between each tagged dev release
19:40 mohawk that is, the substantive overall code changes
19:40 mohawk not each commit message
19:40 mohawk those, from time to time, miss stuff out or are unhelpful
19:40 sivoais I agree 100%
19:40 mohawk ok
19:40 mohawk let me just get the compare URL syntax so i can put it on here for posterity-ish
19:42 mohawk it's pretty technical (especially since they put in the text box top of the drop down, which is amazing) - https://github.com/PDLPorters/​pdl/compare/v2.007...v2.007_01
19:42 mohawk i'm going to do this right now (on a branch, natch)
19:42 sivoais mohawk: just gather all the release notes
19:43 sivoais let me find them
19:43 mohawk that is a good call
19:43 mohawk in fact, that's what belongs in Changes
19:43 sivoais exactly!
19:43 mohawk and in nearly the same format
19:43 mohawk feeling keen?
19:43 mohawk in fact, since it's on a branhc
19:43 mohawk let's be bold
19:43 mohawk remove the "preop" stuff from Makefile.PL (i'll do that)
19:44 mohawk (now, in fact)
19:44 mohawk then you take on the RN reformatting
19:44 mohawk hold on
19:46 mohawk in fact, the preop needs to stay in there for now, since that's how kmx's doc-mangling is happening
19:47 sivoais surely the second line that dumps the git log is what needs to be removed?
19:47 mohawk yes
19:47 mohawk i'm referring to the whole concept ;-)
19:47 opkick [pdl] mohawk2 created changesfix from master (+0 new commits): http://git.io/pxve
19:48 mohawk let's not even PR it yet, since there's no build implications for the present time
19:48 mohawk there you go
19:48 mohawk fire away on that branch
19:48 mohawk wait, i need to make one other change
19:49 sivoais just drop the Changes preop and turn the Release_Notes into into a changelog?
19:49 mohawk i'm also removing the special-case on t/01
19:49 sivoais heh, saw that
19:50 opkick [pdl] mohawk2 pushed 1 new commit to changesfix: http://git.io/pxvH
19:50 opkick pdl/changesfix 63abc7a Ed J: Changes file maintained by person not dumped from git log
19:51 mohawk go :-)
19:52 * mohawk goes back to fighting this MatrixOps thing
19:55 mohawk sivoais, need any further from me on that?
19:55 sivoais mohawk: re: the whole SPLIT_BUILD thing, I was thinking it was working because I saw WITH_HDF => 1 at the end :-P
19:55 mohawk it shouldn't have
19:56 mohawk that sounds like a bug
19:56 sivoais that's very odd then
19:56 sivoais I'm thinking, why not just move Release_Notes to Changes and strip out the install info?
19:56 mohawk effectively yes
19:56 mohawk but it needs to be valid YAML
19:57 sivoais OK, that shouldn't be too hard since YAML is quite flexible :-)
19:57 sivoais it probably is quite close to valid YAML already
19:58 mohawk can you find a yaml syntax-checker?
19:58 sivoais json_pp -f yaml < $INPUT # ?
19:58 mohawk make sure it produces the same as a known-good Changes file
19:59 sivoais like Inline?
19:59 mohawk for instance
19:59 mohawk i'd forgotten about RN
19:59 mohawk i was thinking this would be hard work
19:59 mohawk sivoais++
20:00 travis-ci PDLPorters/pdl#111 (master - 8647162 : Ed J): The build passed.
20:00 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/263d8aeb12ea...8647162a42f6
20:00 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/54388510
20:06 mohawk sivoais, can we take the opportunity (since we're being bold) to also change INSTALL to recommend "cpan"?
20:07 sivoais cpanm?
20:07 mohawk yup
20:07 mohawk this whole manual install thing is a pain
20:11 opkick [pdl] zmughal comment on issue #67: This work is being initiated in the <https://github.com/PDLPorters/pdl/tree/changesfix> branch. http://git.io/pxqe
20:20 opkick [pdl] coveralls comment on issue #21:

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