Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-06-14

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

All times shown according to UTC.

Time Nick Message
00:02 mohawk sivoais, if you want me to i can express that on the list
00:04 sivoais that would be helpful. I am sure seeing one response will lead to others
01:07 opkick [pdl-io-gd] zmughal closed issue #2: Bring in PDL::IO::GD history from PDL repo http://git.io/vIjy5
10:46 mohawk ok
10:46 mohawk i'll make p-c have history first :-)
13:02 opkick [pdl] wchristian created pdl-2x-fixes (+6 new commits): http://git.io/vLv2d
13:02 opkick pdl/pdl-2x-fixes 47f91a3 Chris Marshall: Use $^X instead of "perl" in INCLUDE_COMMAND...
13:02 opkick pdl/pdl-2x-fixes 0c711c2 Chris Marshall: Add File::Path to CONFIGURE_REQUIRES
13:02 opkick pdl/pdl-2x-fixes 8341400 Chris Marshall: Give PDL::NiceSlicee a non-developer version number...
13:02 opkick [pdl] wchristian tagged v2.012 at 47fed3d: http://git.io/vLv2F
13:09 mohawk sivoais, chm is alive! ;-)
13:09 mohawk ...and his branch is 13 commits behind master
13:10 mohawk i'm sending him an urgent email as it doesn't merge and it's breaking perl-in-space
13:13 mohawk oh dear god
13:14 mohawk why the hell has he released to cpan from a branch?
13:18 chm joined #pdl
13:18 mohawk i'm emailing now, with cc sivoais
13:18 mohawk ah, chris
13:18 mohawk i'm just emailing you now
13:18 mohawk there are problems
13:18 mohawk sent
13:19 chm hi, just saw the release, I was going to send an email but since you're active...
13:19 mohawk sorry, you "saw the release"? didn't you make it?
13:19 chm I set up a branch for finish the PDL-2.x release process
13:20 mohawk i'm having a bit of a headdesking feeling, because this isn't an ideal process
13:20 chm It was a case where the versions for 2.008, 2.009, 2.010, and 2.011 all had different issues introduced as something got fixed.
13:20 mohawk please let me know when you've seen & read my email
13:20 mohawk well, that's the nature of the beast
13:21 mohawk and you've been nailing the issues and releasing updated, improving versions?
13:21 chm If you'll take a look, the changes are the least I could do to get PDL-2.x stable working
13:21 chm There were some missing packages and a regression from a perl-in-space fix that didn't work correctly
13:22 mohawk what makes you say it didn't work properly?
13:22 chm Just got your email, let me read and pop bakc...
13:22 mohawk great
13:23 mohawk point 1 is particularly critical, and if you ok it, i can fix it up while we chat about other stuff
13:24 chm Is the problem that the pdl-2x-fix branch is trying to be forced into master?
13:24 mohawk it's worse
13:24 mohawk you released to CPAN
13:24 mohawk my vision is that master is what's gone to cpan at various points
13:24 chm That vision doesn't work if master moves on while CPAN is broken
13:25 mohawk there is no point at all in branch-based dev if there's no review
13:25 mohawk you've just unilaterally decided that master is no good, so you'll just make a new stable-ish branch
13:26 mohawk that's a model people use, but it's a new model and there was no discussion
13:26 chm No, I've had reports that there are indexing failures in the only extant PDL official release.
13:26 mohawk yes
13:26 chm Unfortunately, those issues cannot be detected or fixed without CPAN releases.
13:26 mohawk i saw those reports, and sivoais made a ticket for that
13:27 chm I am also concerned that a large number of ASperl platforms including *all* Mac OS X failed to build.
13:27 mohawk so am i right you're using the CPAN official release as a pseudo-dev release but engaging the indexing process?
13:27 mohawk i emailed a contact at activestate and am waiting for him to get back
13:28 mohawk that might take a day or 3
13:28 chm Since the release is out, it should be testable.  If there is a better way to "debug" CPAN indexer problems, I'm open
13:28 mohawk believe it or not, the cpan indexer "trying things" issue is one i started working on about a year ago
13:29 chm Part of the difficulty is that we've made a large number of non-trivial changes to the build process.
13:29 chm A lot of those directly affect the indexing process.
13:29 mohawk PAUSE is a gigantic monolith of code, and it's not yet possible to take bits out and use them independently
13:30 chm Unfortunately, I was pretty psyched about the new release and its patina, that I deleted PDL-2.006 and PDL-2.007
13:30 mohawk they're equally available in backpan, nothing changed
13:30 chm The result was that some legacy indexer problems hidden behind the old release now popped up.
13:30 mohawk the way to beat these indexing problems is to do what sivoais and i are finishing up - to break up the iceberg
13:31 chm For example, having my work to improve PDL::NiceSlice with a developer version seemed reasonable.
13:31 mohawk please could you educate me on why you think the INCLUDE_COMMAND -> perl instead of $^X isn't "viable"?
13:31 chm Then when the earlier releases disappeared, so did PDL::NiceSlice (as far as the indexer is concerned)
13:32 mohawk having independent versioning within PDL isn't really a viable model
13:33 chm I agree, it is a mess and moving away is a good thing.  However, leaving a serious regression for the PDL stable release is unacceptable.
13:33 mohawk ok
13:33 chm It seemed to be the least damaging way to address the issue.
13:33 mohawk a bit of discussion might have been a good way forward
13:34 mohawk you've announced 2.012 as being from the team
13:34 mohawk well, "the pdl developers"
13:34 chm It is as the 4 lines from me plus the zillion lines from the team...
13:34 chm I'm proud of the work that has been done.
13:34 mohawk well you've made a release of PDL that breaks on "perl in space"
13:35 mohawk to say i find that disappointing would be an understatement
13:35 chm You'll have to update me on what "perl in space" is as I have no clue.  What ticket is it fixing?
13:35 chm I never did understand the git log entry.
13:35 mohawk ok
13:36 mohawk if i make code changes you don't understand, perhaps asking questions would be useful :-)
13:36 mohawk it's simply perl installed in a directory with a space in its name
13:37 mohawk unfortunately EU::ParseXS $^X in an INCLUDE_COMMAND doesn't work right with such a scenario
13:38 chm Re $^X, in ExtUtils::ParseXS you have this a few lines above the INCLUDE_COMMAND:     Warn( $self, "The INCLUDE directive with a command is discouraged." .
13:38 chm " Use INCLUDE_COMMAND instead! In particular using 'perl'" .
13:38 chm " in an 'INCLUDE: ... |' directive is not guaranteed to pick" .
13:38 chm " up the correct perl. The INCLUDE_COMMAND directive allows" .
13:38 chm " the use of \$^X as the currently running perl, see" .
13:38 chm " 'perldoc perlxs' for details.");
13:38 mohawk i know
13:38 mohawk it's a real pity it doesn't work right
13:38 chm In fact, I believe that to be the problem with ASperl
13:38 mohawk that's entirely possible
13:38 chm And yes, it is a huge PITA
13:39 mohawk a better solution would be to just generate .c/.xs files while we wait for EU::PXS to work
13:39 chm The sense that the change was a regression was that PDL *never* worked with whitespace paths fully
13:39 mohawk well it actually did
13:39 mohawk 2.011, for a start
13:40 chm That's news to me since there were always edge cases that were not guaranteed
13:40 mohawk 2.012 definitely won't
13:40 mohawk the generated-files would still work on perl-in-space
13:40 chm Do you test your whitespace fix with two different perls, $^X and "perl" to see if you get the observed problem
13:40 mohawk yes
13:41 mohawk that was how i knew the "perl" option definitely worked
13:41 mohawk in fact, more accurately, on win32 i only use "perl in space"
13:41 mohawk anyway
13:41 mohawk right now i'd like to switch master and your branch (just the labels) so we still have master = released
13:41 chm We'll a fix for win32 that breaks MacOSX is not a complete solution
13:42 mohawk by "macosx" do you mean the ASPerl builds?
13:42 chm Yes.  I personally use cygwin or strawberry but there are a lot of folks using ASperl "by default"
13:42 mohawk ok
13:43 mohawk so it's not breaking on the general macosx case
13:43 chm I hope that this PDL release can be the first one with *full* builds on ASperl
13:43 chm Something to go in the "done basket"
13:43 mohawk if you want ASperl building ok, the way to achieve that is via communication with AS
13:43 mohawk this is something i am doing as it's worked great in the past
13:44 mohawk now
13:44 mohawk would you like me to switch the branch labels?
13:44 chm What I want is PDL building out-of-the-box and we have talked with ASperl as well.
13:44 chm However, if we change something in our release that causes the PDL release following 2.007
13:45 chm to *not* *build* on *many* *perls* then we should fix that on our end and not jigger things at ASperl
13:45 mohawk in case it's not clear, i'm committed to PDL building everywhere, including AS
13:45 chm That said, as Rob has mentioned, it is perfectly reasonable not to support old, crufty MSVC going forwards either
13:46 mohawk please could you answer the branch-switching question
13:46 mohawk i'm going to keep asking it till you do
13:47 chm I think that is a good idea.  Maybe we could have held off on the master updates until PDL-2.x was stable again
13:47 chm Maybe a bug fix for the little things as well.
13:47 mohawk well, releasing things without discussion has made effectively a master update but not acknowledging that
13:47 mohawk and i do think you are 100% correct on the necessity of making full releases in order to see what the indexer does
13:48 mohawk but... discussion
13:48 mohawk doing the switching now
13:48 chm I had seen a number of changes to master that would have made a "quick re-release" difficult
13:48 mohawk can you outline which ones?
13:50 chm The change to Changes for one.
13:50 chm While that may be cosmetic for you, it is a big change in where release information goes
13:50 mohawk oh no
13:50 mohawk please tell me you didn't release from behind that
13:51 chm I released from v2.011
13:51 chm see the pdl-2x-fixes branch graph
13:51 chm literally 4 lines or so of changes plus release notes, known problems,...
13:52 mohawk so 2.011 is pre the changes change?
13:52 chm yes, v2.008 through v2.012 are all the original PDL-2.x lineage
13:53 chm The plan was to get a stable release out and then move on
13:53 chm The catch was the index and build problems that took a while to pop up
13:53 mohawk please can you help me understand why the changes change makes a quick release difficult?
13:54 chm "The plan was to get a stable release out and then move on"
13:54 chm The first thing in the move on was new changes, etc.
13:54 mohawk i see
13:54 chm The out-of-order execution is from the indexer/build problems that took weeks to unfold
13:54 mohawk so a kind of step-by-step movement idea?
13:55 chm This would not be as critical except that I over eagerly deleted PDL-2.007 from CPAN
13:55 opkick [pdl] mohawk2 created unmerged-to-master from pdl-2x-fixes (+0 new commits): http://git.io/vLvDS
13:55 mohawk oops
13:55 chm That left PDL-2.0xx holding the bag and dropping the ball a bit
13:55 mohawk well, fair enough
13:55 mohawk we will need to capture all the lessons learned into the DEVELOPMENT doc
13:55 mohawk that is a tad out of date
13:56 chm I agree.
13:56 mohawk let's not delete main releases until at least a month has elapsed after the next one :-)
13:56 chm It probably would be a good idea to hold of master changes for that period as well.
13:56 opkick [pdl] mohawk2 force-pushed master from a96c184 to e01c825: http://git.io/F_ij-Q
13:57 mohawk i think that would be a less good idea
13:57 mohawk maybe a few days or a week for such a "purdah"
13:57 mohawk we are in a period of instability which will pass
13:58 chm Maybe a week is the right balance.  Also, if we leave a working stable release in the CPAN we're covered that way.
13:58 mohawk i was already figuring how to fix the PDL::Lite etc issues for pdl-core
13:58 mohawk that's my thinking
13:58 mohawk glad we're on the same page :-)
13:58 mohawk so, now master is the released version and we have a not-yet-merged branch
13:58 chm I think if we generally assume that we are all trying to do the right thing for PDL
13:59 mohawk i'm going to delete your "2x-fixes" branch since it's now not relevant
13:59 chm and go from there, things'll work out.  I definitely am not perfect but I am trying
13:59 mohawk i honestly think the process we're still moving out of, where everything is in one boat, has led to a lot of unnecessary tension
13:59 chm to help PDL to evolve as a project while being increasingly usable to the PDL community
13:59 mohawk because the risk in a release is so great
14:00 mohawk chris, you resurrected PDL from its doldrums
14:00 mohawk more recently, you moved it to using git which has helped a lot
14:00 chm I agree, but again, we need to have a *stable* PDL on CPAN while we push the boundaries
14:00 chm Personally, I would prefer that we fix the 64bit index support as the #1 priority
14:00 mohawk even more recently, you're embracing technologies and techniques that will move things forward at new levels of speed and correctness
14:01 mohawk the history of PDL in the last 10 years will basically be your vision
14:01 chm That is a key capabiity that we are close to having but I've been unable to debug myself.
14:01 mohawk i think it's different
14:01 mohawk having PDL::Core with so much stuff around it greatly (GREATLY) slows the dev process
14:01 mohawk we MUST move to split model first
14:02 mohawk then we can fixup the 64bit thing
14:02 mohawk we could have core split out in a week
14:02 mohawk that is literally true
14:02 mohawk all we're waiting for is your blessing
14:03 chm Is all the "splitting out core" in one branch?
14:03 mohawk it's in a branch now
14:03 opkick [pdl] mohawk2 deleted pdl-2x-fixes at e01c825: http://git.io/vLv9T
14:03 mohawk but it's going to get put in its own repo today
14:03 chm What were your plans for namespace handling?
14:03 mohawk good question - but i don't understand it?
14:04 chm Can we start with a version that could be swapped in for the stable PDL-2.x Core?
14:04 mohawk the blessing required would be either a release from you, or comaint for sivoais
14:04 mohawk since it will be literally copied over (including the history), it will be that by definition
14:05 mohawk ah, i think i understand - the plan is to just release it as PDL::Core, so same namespace
14:05 chm Lets get PDL-2.0xx resolved first.  That way the appearance of a new release won't break things
14:05 mohawk so there would be a release of PDL::Core 2.013 (say), and PDL 2.013 without P::Core in it
14:05 chm says someone recently burned by official vs developer release differences :-)
14:05 mohawk indeed
14:05 mohawk well, there will be a few more teething troubles but we will move quickly and fix them
14:06 mohawk we can make a best effort to minimise problems
14:06 chm More likely there could be PDL-2.012 with Core the works with PDL::Core as well
14:06 mohawk no, we can't
14:06 chm Then PDL-2.013 could be separate?
14:06 mohawk because there's already a PDL::Core 2.012
14:07 chm Version issues?  Those are things that I would like to understand better before an official split
14:07 mohawk there isn't much to understand
14:07 mohawk you can't release the same version of something that's already released
14:08 mohawk there's already effectively a PDL::Core 2.012
14:09 chm We could do the PDL-2.013 release with PDL::Core no-indexed and have the rest in PDL::Core but indexed
14:09 mohawk this process we are looking at is a risk-reducing strategy, not a risk-increasing one
14:09 chm That's what I would like to ensure
14:09 mohawk chris, when you release software there is risk
14:10 mohawk if you can't take the heat, the kitchen is no place to be :-)
14:11 chm ed, please re-read the previous discussion as regards my concerns about having a stable, working PDL out
14:11 mohawk the thing you're talking about for 2.013 is the same as i am, except that having P::C no-indexed in main release is a bad idea that will cause clashes or mask them
14:11 mohawk i want the same thing
14:11 mohawk there is no risk-free option
14:11 mohawk all software has bugs
14:11 chm I would hope that we could plan thing a bit better than #pdl banter before taking action.
14:11 mohawk just not releasing any more versions means those old bugs will eventuate
14:12 mohawk and reported ones aren't fixed in the released one
14:12 mohawk zaki wrote a plan and emailed it yesterday
14:12 chm I haven't caught up on my mail yet---mostly debugging PDL-2.012 stuff and this conversation
14:12 mohawk are you saying plans don't exist unless you write them? :-)
14:13 mohawk my email is a little behind too, so i get that ;-)
14:13 chm I'm saying with more than 1 person in a team, coordination is best when it is very clear what the plan is
14:14 chm My experience with #pdl is that there is a lot of dialog but something like "meeting notes" are needed to codify things better
14:14 mohawk so far they've always been reflected in emails
14:15 chm Great catching up with you, breakfast calls, plan to catch up on email this PM.  o/
14:15 mohawk nothing's been said here today that's not in zaki's email
14:15 mohawk i have another 7-8 hours online so we'll have to chat (or email) during that window
14:15 mohawk ciao ;-)
14:42 opkick [pdl-io-gd] devel-chm comment on issue #3: A key issue for modules that have been split out from the PDL
14:51 travis-ci PDLPorters/pdl#248 (master - e01c825 : Chris Marshall): The build passed.
14:51 travis-ci Change view : https://github.com/PDLPorters/pdl/compare/a96c18403ecf...e01c825bb0d6
14:51 travis-ci Build details : http://travis-ci.org/PDLPorters/pdl/builds/66748525
15:15 opkick [pdl-io-gd] mohawk2 comment on issue #3: This might need a rethink of how the pdldoc stuff works. http://git.io/vLfTm
18:41 opkick [pdl] zmughal opened pull request #121: New changelog format (master...changesfix) http://git.io/vLJeS
18:49 opkick [pdl-io-gd] zmughal comment on issue #3: Very good points. We'll need to add a test to make sure that this always works no matter how we may build the pdldoc database in the future. Better to check breakage first in dev.... http://git.io/vLJfS
18:49 opkick [pdl-io-gd] zmughal reopened issue #3: pdl-io-gd split tasks http://git.io/vIQSD
18:52 opkick [devops] zmughal pushed 1 new commit to master: http://git.io/vLJJp
18:52 opkick devops/master 1b8d625 Zakariyya Mughal: add a task to ensure that `pdldoc` indexing works after split...
18:57 opkick [pdl-io-gd] devel-chm comment on issue #3: A standard hook sounds like the right direction.  I'm with Ed that we might
19:03 opkick [pdl-io-gd] zmughal comment on issue #3: I was looking online for past discussion and saw that you [commented](http://comments.gmane.org/gmane.comp.lang.perl.pdl.general/6272) that ... http://git.io/vLJkE
19:08 opkick [pdl] zmughal closed issue #120: SF#389 not indexed modules http://git.io/vIjUF
19:21 opkick [pdl-io-gd] mohawk2 comment on issue #3: I think that add_doc.pl is very similar to the code in PDL::Doc's "EXAMPLE". What this tells me is we just need an extra bit of API to encapsulate this. Shouldn't be hard. Thoughts? http://git.io/vLJmG
19:23 opkick [pdl-io-gd] mohawk2 comment on issue #3: Re @zmughal 's quote of a previous remark by @devel-chm , let's not get overcomplicated. I think adding docs that can then be found in the same way as now, i.e. preserving current functionality under the new scheme is enough. What do you guys think? http://git.io/vLJmo
19:28 opkick [pdl-io-gd] devel-chm comment on issue #3: I think keeping it minimal---just enough to maintain the pdldoc status quo
19:30 opkick [pdl-io-gd] zmughal comment on issue #3: Right, doing it the same way will get the release out the door faster without backcompat issues.... http://git.io/vLJOL
19:57 opkick [pdl-io-gd] mohawk2 comment on issue #3: I assume you mean PDL::Doc, not PDL::Core? It will work in the same way as currently, but hide the detail that is currently revealed/required. http://git.io/vLJZ3
19:57 opkick [pdl-core] zmughal opened issue #2: Look into a standard hook for updating pdldoc database http://git.io/vLJZZ
19:57 mohawk sivoais, that's still a pdl not a pdl-core issue :-)
19:59 opkick [pdl-io-gd] zmughal comment on issue #3: Yes, @mohawk2. See the issue I created here <https://github.com/PDLPorters/pdl-core/issues/2>.... http://git.io/vLJZ6
20:00 sivoais mohawk: it's not? or will it not be?
20:00 sivoais are we planning on creating a separate ::Doc dist?
20:00 mohawk i would say it seems a prereq for moving to p-c
20:00 sivoais hmm
20:01 mohawk i'm thinking of just encapsulating what add_doc.pl (and the P::Doc example) does into a method
20:01 mohawk eg find_dbfile
20:02 sivoais how about add_tree( ) to match with add_module( ) ?
20:02 * sivoais looks at Doc.pm
20:03 sivoais yeah, pull out the directory searching code in add_module() to its own function
20:03 mohawk how about just having $pd = PDL::Doc->new($file_you_specify_yourself) replaced with $pd = PDL::Doc->new(PDL::Doc->find_dbfile)
20:03 mohawk or whichever
20:03 sivoais yes
20:04 mohawk garg, add_module is a function not a method
20:04 mohawk #needsfixingnow
20:04 mohawk anyway, that bit of code will come out of there into a class method
20:05 sivoais well, add_module is meant to be called in builds
20:05 mohawk hm
20:05 mohawk well in that case, let's just use that
20:05 mohawk so does the problem we're facing just boil down to PIG isn't calling that yet?
20:05 sivoais yes, that's all ;-)
20:05 mohawk and if so, isn't that just fixed by adding to PDL::Core::Dev::pdlpp_stdargs[_int]?
20:07 mohawk or possibly postamble
20:08 sivoais yes, postamble
20:08 mohawk hm, not entirely
20:08 mohawk what's the EUMM hook for install-time stuff?
20:09 sivoais <https://metacpan.org/source/DKOGAN/PDL-FFTW3-0.02.2/Makefile.PL#L91>
20:09 sivoais that uses MY::postamble
20:09 sivoais and <https://metacpan.org/source/DCMERTENS/PDL-Parallel-threads-0.03/Build.PL#L12>
20:09 sivoais uses ACTION_install
20:09 mohawk a :: type, fine
20:10 mohawk please don't sully this channel with your references to Build.PL
20:10 mohawk ( ;-) )
20:10 sivoais hahah
20:10 sivoais when was the last time run4flat came in here?
20:10 mohawk ages i think
20:10 mohawk ask jberger :-)
20:11 mohawk so yes, i can just steal that bit of code for P::C::Dev::pdlpp_postamble itself
20:12 sivoais mohawk: could you make it more clean while you're at it?
20:12 sivoais because add_module finds the directory AND adds it
20:13 sivoais and the only difference between add_module and add_doc.pl (from PDL-Stats) is that add_doc.pl calls scantree() instead of scan()
20:13 sivoais or do you want me to do it?
20:13 mohawk back-compat
20:14 sivoais let me make a PR to show you
20:14 mohawk are you proposing an api change?
20:14 sivoais no, an additional class method
20:14 mohawk sure, if it doesn't take much effort
20:14 mohawk but what will it achieve that can't already be done?
20:16 sivoais hmm, that's true
20:16 mohawk let's not agonise about minor infelicities
20:16 sivoais if you add it the the standard pdlp_postamble, then yes, everyone would benefit immediately
20:17 mohawk exactamundo
20:17 mohawk soon as i've eaten, it's getting PRed
20:25 mohawk sivoais, if you look at that function, i'm going to make add_module be called with PDL::$mod
20:25 mohawk thoughts?
20:26 sivoais one sec, add pdlpp_postamble?
20:26 mohawk called from within that
20:26 sivoais q_q rebasing against this new master is causing merge conflicts
20:26 mohawk my($src,$pref,$mod) = @$_;
20:27 mohawk i know
20:27 mohawk feel free to be aggressive in fixing
20:27 mohawk install ::
20:27 mohawk \@echo "Updating PDL documentation database...";
20:27 mohawk \$(PERL) -e 'exit if \$\$ENV{DESTDIR}; use PDL::Doc; eval { PDL::Doc::add_module(q{PDL::$mod}); }; ';
20:30 mohawk this is the changes from today's commits: https://github.com/PDLPorters/pdl/compare/fa428b0826162cc56e0f23215338115336027daf...master
20:34 sivoais mohawk: shouldn't $mod contain PDL:: already?
20:35 sivoais if you look at how both pdlpp_postamble{,_int} are called, the $mod contains the full package name
20:36 mohawk excellent
20:36 mohawk good research
20:38 mohawk just running tests now...
20:38 sivoais odd
20:38 sivoais is this right?
20:38 mohawk ?
20:38 sivoais so in PDL-IO-GD's Makefile.PL, pdlpp_postamble( 'PDL::IO::GD' ) is the call
20:39 sivoais while in many other places, there are more arguments
20:39 mohawk i don't think that's right, then
20:39 mohawk supposed to be arrayref apart from anything else
20:40 sivoais yeah, grep the main pdl source for use of pdlpp_postamble, you'll see the same <https://github.com/PDLPorters/pdl-io-gd/blob/master/Makefile.PL#L104>
20:40 sivoais in other places
20:41 mohawk what's the value of $package?
20:41 sivoais IO/GD/Makefile.PL, IO/HDF/SD/Makefile.PL, IO/HDF/VS/Makefile.PL, Lib/GIS/Proj/Makefile.PL, Lib/Transform/Proj4/Makefile.PL
20:41 sivoais "PDL::IO::GD"
20:41 mohawk in PIG i mean
20:41 mohawk read it
20:41 sivoais ah...
20:41 sivoais wow
20:41 sivoais nooooo
20:41 sivoais ok, I see
20:41 sivoais nvm
20:42 mohawk 200 pressups
20:42 mohawk ;-)
20:42 sivoais noooo... well, I do need the exercise
20:42 mohawk ha ha
20:43 mohawk what other tasks are there apart from this doc thing?
20:43 mohawk rebase finished ok?
20:43 mohawk could you email the list telling them we had to force-push master?
20:45 sivoais yes, also, i think using -Xtheirs for the merge strategy would make sense as long as people have been keeping up to date
20:45 mohawk uh, what's that?
20:45 mohawk their master should all be just tracking the SF one?
20:46 sivoais git rebase -Xtheirs origin/master
20:46 mohawk which does what?
20:46 mohawk (hands busy eating :-)
20:46 mohawk (just installing my local PDL so i can try PIG against it)
20:47 sivoais well, the rebase will hit a merge conflict and when it does, it will chose what is on master over what is in the current branch
20:47 sivoais for me, that is the fact that 2.011_01 went away and became 2.012
20:48 sivoais we had code for 2.011_01, right?
20:48 mohawk yes
20:48 mohawk i switched it to unmerged-to-master
20:48 mohawk i really just switched the labels around
20:49 sivoais yeah
20:49 mohawk so it's still there
20:49 mohawk but it will need rebasing
20:49 sivoais but now it is not and that screws up the merging
20:49 sivoais errr, the rebasing
20:49 mohawk uh, what?
20:49 mohawk you should always rebase -i
20:50 mohawk some human judgement required
20:50 mohawk -X not really called for
20:50 mohawk and i doubt our pdl comrades are doing much rebasing
20:50 sivoais git log master | grep 2.011_01 # nothing
20:50 sivoais git log test-cleanup-3 | grep 2.011_01 # Update VERION to 2.011_01 for more development
20:51 sivoais there was an intermediate version and now it is gone
20:51 mohawk fine
20:51 mohawk it's not gone
20:51 mohawk it's still on unmerged-to-master
20:51 mohawk comprende?
20:52 mohawk hold on
20:52 mohawk i think i have messed up
20:52 sivoais git log origin/unmerged-to-master  | grep 2.011_01 # nothing
20:52 mohawk correct
20:52 mohawk let me retrieve it from my reflog
20:52 mohawk unless you have SHA of previous master HEAD?
20:54 sivoais a96c18403ecfeb8af9f15cb1b8d55e9daa8742ed
20:54 sivoais right?
20:54 mohawk yup, i agree
20:56 opkick [pdl] mohawk2 force-pushed unmerged-to-master from e01c825 to a96c184: http://git.io/vLvDS
20:56 mohawk that should be it
20:57 mohawk looks like some pressups for me too ;-)
20:57 mohawk confirm it's what you expect?
20:59 sivoais yep, lgtm
20:59 mohawk groovy
20:59 mohawk if you could rebase minus the 11_01 that would be smashing
21:02 sivoais on to the master? or create a new branch?
21:03 sivoais I'll create a new branch and then you can check it for sanity
21:03 mohawk ok
21:03 mohawk this new doc thing looks good
21:04 mohawk going to PR on pdl
21:05 opkick [pdl] mohawk2 created pdldocfix (+1 new commit): http://git.io/vLJuv
21:05 opkick pdl/pdldocfix d484cfd Ed J: Add PDL::Doc::add_module to pdlpp_postamble
21:07 opkick [pdl] mohawk2 opened pull request #122: Add PDL::Doc::add_module to pdlpp_postamble (master...pdldocfix) http://git.io/vLJuG
21:07 mohawk we are living in an age of magic
21:08 mohawk i added this snippet in PDL's pdlpp_*, ran it against my local PIG
21:08 mohawk it just worked
21:08 mohawk now i push the change to GH, and PR it
21:08 mohawk that magically appears in this irc channel
21:09 mohawk meanwhile, it's all getting comprehensively CI-ed against all the perl versions we support
21:18 opkick [pdl-io-gd] mohawk2 comment on issue #3: A suitably minimal (4-line) fix that makes PDL::IO::GD "just work" in pdldoc: https://github.com/PDLPorters/pdl/pull/122 - though it will need a suitably up-to-date PDL once we release this. http://git.io/vLJg1
21:27 mohawk sivoais, i made an issue on FFTW3: https://github.com/dkogan/PDL-FFTW3/issues/5
21:29 sivoais mohawk: one thing
21:29 sivoais i'll note it on the issue
21:31 opkick [pdl] zmughal comment on issue #122: For dists like [PDL-Stats](https://github.com/maggiexyz/PDL-Stats/blob/master/add_doc.pl) and [PDL-Parallel-threads](https://github.com/run4flat/PDL-Parallel-threads/blob/master/Build.PL#L13) which use `scantree`, how will they be affected? http://git.io/vLJar
21:31 mohawk good question
21:31 mohawk no idea
21:31 opkick [pdl] zmughal created unmerged-to-master-drop-2.011_01 (+11 new commits): http://git.io/vLJaD
21:31 opkick pdl/unmerged-to-master-drop-2.011_01 6fe9c8d Derek Lamb: Let PDL::DiskCache::sync accept an argument of which element to sync....
21:31 opkick pdl/unmerged-to-master-drop-2.011_01 0be7f21 Derek Lamb: 2 small cleanups in slices.pd...
21:31 opkick pdl/unmerged-to-master-drop-2.011_01 5ce402a Derek Lamb: Add curly braces to remove -Wdangline-else...
21:32 sivoais that's why I was suggesting creating a new class method
21:32 mohawk i just filed an issue on P-Stats: https://rt.cpan.org/Ticket/Display.html?id=105227
21:32 mohawk i severely doubt it will have any material effect
21:32 mohawk if it does, we can fix it
21:33 mohawk let us not forget "YAGNI", who died for our overengineering sins
21:33 sivoais hmm, yeah, adding to the DB should be idempotent
21:33 mohawk all hail YAGNI
21:33 sivoais never forget
21:33 mohawk ha ha
21:34 mohawk author of P::Stats: https://www.youtube.com/watch?v=DFX_cNB97yQ
21:34 sivoais ok, so what I did was piece together the commits added to the pre-2.012 master
21:34 sivoais and put them on top of the current master
21:34 sivoais check unmerged-to-master-drop-2.011_01
21:34 mohawk just did
21:35 mohawk i thought there were 13 commits past 2.011, which minus the VERSION update = 12, not 11?
21:35 sivoais I dropped the very last one
21:35 sivoais as well
21:36 mohawk fair one
21:36 mohawk so that looks fine
21:36 mohawk might as well PR it?
21:36 mohawk just to engage the CI magic
21:37 sivoais yep, done
21:37 opkick [pdl] zmughal opened pull request #123: Fix for force-push post-2.012 (master...unmerged-to-master-drop-2.011_01) http://git.io/vLJV9
21:39 sivoais *phew* ... I'm glad I know about git format-patch and git am... otherwise working with that is tricky to reason about
21:39 sivoais because of how the Changes file patches don't apply cleanly while it goes through
21:41 mohawk yowzer
21:44 sivoais ok, I'm going to be afk, but once that works, that can be pulled on to master and then we can tell the list about the rebase on master?
21:44 sivoais let me check against my test-cleanup-3 branch too
21:46 sivoais git checkout test-cleanup-3; git rebase -i unmerged-to-master-drop-2.011_01; [ delete the 2.011_01 line, save, quit ]
21:46 sivoais force push
21:50 sivoais ok, I'm going to be afk until later
21:52 mohawk ok
21:53 mohawk my plan is instead to keep these things unmerged to master
21:53 mohawk until we're sure 2.012 is good
21:53 sivoais sounds fine
21:54 sivoais I'll add a note to the the pul
21:54 sivoais l
21:54 mohawk so if you can make sure you're happy with tc3, rebased against current master, that will be useful
21:54 mohawk great
21:54 mohawk i need to figure out a different fix for the INCLUDE_COMMAND that will work on perl-in-space
21:55 mohawk and i need to wade back into the snake's nest that is p5p to get a fix for EU::PXS merged/released
21:56 mohawk sivoais, may i suggest fixup-ing the last 2 commits into one?
21:57 mohawk otherwise between those two travis would fail?
21:57 sivoais ah, sure
21:58 opkick [pdl] zmughal force-pushed test-cleanup-3 from e2cf894 to 82f83a6: http://git.io/vT7A9
21:58 opkick pdl/test-cleanup-3 e1b4b9f Zakariyya Mughal: test cleanup: t/bool.t
21:58 opkick pdl/test-cleanup-3 8615715 Zakariyya Mughal: test cleanup: t/clump.t
21:58 opkick pdl/test-cleanup-3 3977ebe Zakariyya Mughal: test cleanup: t/familyfree.t
21:59 mohawk nice
21:59 mohawk i'm having a go at EU::PXS now...
21:59 mohawk could you announce the force-push?
22:00 opkick [pdl] zmughal force-pushed unmerged-to-master-drop-2.011_01 from 2bf19b3 to 3ca9250: http://git.io/vLJKR
22:00 opkick pdl/unmerged-to-master-drop-2.011_01 3ca9250 Zakariyya Mughal: add author test to check Changes log and run in Travis-CI...
22:07 opkick [pdl] zmughal comment on issue #121: This PR will be obsoleted by #123 when that is merged. http://git.io/vLJ6c
22:17 sivoais here's the e-mail <http://permalink.gmane.org/gmane.comp.lang.perl.pdl.devel/6540>
22:18 mohawk reading
22:19 mohawk looks great
22:19 mohawk nice work!
22:20 mohawk you closed #121
22:20 mohawk does #123 have derek's changes?
22:21 mohawk i see that it does
22:21 mohawk nothing to see here
22:21 mohawk go about your business
22:27 sivoais mohawk: could you send an e-mail about the pdldoc PR to the list too? That way it becomes clear to everyone what the changes are for?
22:27 sivoais right now it's just chm, you, and me that are in the loop
22:28 mohawk i will
22:28 sivoais as well as the couple bug reports you sent :-) Good work on that, mohawk++
22:30 sivoais once that and the indexing troubles are squared away, we just need to check that the whole system works with a split-off PDL-Core and I think we're ready to move ahead!
22:30 mohawk thanks ;-)
22:30 mohawk i think you're right
22:31 sivoais all right then, I'm off o/ later!
22:33 mohawk later amigo

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