#foswiki 2014-03-24,Mon

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
pharveythere are some funking things you could do with the github API too but I don't want to advocate too many bikesheds :) [00:00]
gac410Well, it could be right in the Extension form - point to the repo [00:00]
pharveythat would be it, I think that thought has crossed my mind before. [00:00]
gac410So Tasks tracks the Foswiki project. Foswiki project uses user foswiki on gh. "Developed in SVN" becomes "Developed on Foswiki" (meaning the gh user.
foswiki/foswiki becomes the core+default. foswiki/* the non-default extensions.
I guess my REV2REF has to expand to /repo/######## to support the repo-per-extension
[00:02]
pharveyperhaps it should Foswikiref:SomePlugin/deadbeef
rather than Foswikiref:repo/deadbeef
I seem to recall wanting to add a few fields to the Extensions DataForm
[00:06]
gac410hm... Well the repo is SomePlugin today. [00:07]
pharveyso REV2REF would consult SomePlugin's dataform, build the URL appropriately, assuming it knows about that plugin's commit history [00:07]
gac410Actuallly Foswikiref:xxxx is whatever we want, and my macro isn't needed. Foswikirev:nnnnn is what the macro changes to a hashref
Once we "convert" new commits get written directly as Foswikiref links, It's the Foswikirev links that need translation.
does that make sense?
[00:08]
pharveytrue
sorry if I'm a bit distracted, debugging a protocol bug in something that abuses telnet RFC2217 :P
pharvey hugs wireshark
[00:09]
gac410gac410 says 5 min. with wireshark teaches you more about protocols than an hour with a book, [00:10]
pharveyyep
I've been using it since it was called ethereal
silly trademarks
[00:10]
gac410I'll try to summarize this discussion into the Migrate task. Yeah same here. I was a multiprotocol router guy from the early days. [00:11]
So I'm not sure there is any good reason for submodules within our project. Core+default is monolithic. Each non-core extension is separate repo. pseudo-install can check them out regardless of sub-module or separate module status. [00:18]
Babargac410: pharvey: only read the very beginning, but the github "master" which pushes to github is on the server. Feel free to clone it locally (still on the server), and play with it [00:18]
gac410I think I understand what's going on mostly. [00:19]
Babargood then
shall I upgrade the server btw?
Babar read that gmc did some of it
[00:19]
gac410gmc was working on it. pootle is kapootled ... wsgi issues.
gac410 wondering if we should evaluate moving to http://weblate.org. It seems to have excellent github support, including cross-branch translation, which was always an issue with pootle.
[00:20]
Babaroh man... FreeBSD switched to binary packages! [00:21]
gac410gac410 doesn't do translation though, so really has no idea. [00:21]
BabarBabar tried pootle for a while. Didn't like it [00:22]
gac410I had to compile svn. Managed to switch http support back on and recompile it. [00:22]
Babarwhy would we need http support in svn?
oh, for its local clones?
[00:22]
gac410the local svn client couldn't check out
right.
[00:22]
Babarwe could have used the path :) [00:23]
gac410trunk stopped updating. http://.... url not understood,. [00:23]
Babarbut that would have been shitty [00:23]
gac410anyway, it's working. I was lucky. [00:23]
Babarno. You're just getting good at it. Not lucky :) [00:24]
gac410Don't count on that. [00:24]
BabarI do. OK, I'm hungry. I'll go grab some healthy American food at the local 3 stars restaurant and I'll be back [00:25]
gac410:) [00:25]
Babarstarted the upgrade to pkgng. I'll finish when I'm back
shouldn't do anything, it's just all internals
[00:29]
gac410Okay. great. [00:30]
pharveyhah! they forgot to convert decimal -> hex. Lame [00:32]
...... (idle for 28mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/B1sNNg
foswiki/master 4b1a1a7 GeorgeClark: Item12821: Initial commit map svn rev -> git hash...
[01:00]
***GithubBot has left [01:00]
FoswikiBothttp://foswiki.org/Tasks/Item12821 [ Item12821: Create FoswikirefsPlugin to map svn revisions to git hashrefs ] [01:00]
............ (idle for 56mn)
Babarupgrading the server. I've briefly scanned through the upgrade list, and it should be fine.
I'll double-check once done
but as it no longer needs to re-compile the entire world, it should be done fairly quickly
[01:56]
gac410cool
translate.f.o is still dead.
gmc said he was running into a lot of dependency issues.
I had a huge issue getting stuff running after the perl upgrade. I don't recall the details now, but I just could not get cpan to install the dependencies for the bots.
[01:57]
Babarok, I'll fix it
or at least I'll try
Upgrades have been requested for the following 603 packages:
[02:03]
gac410yeah it was a bit back level :) [02:06]
Babargreat... connecting to foswiki.org hangs at... connecting to yui.yahooapis.com
(from my laptop)
(which is like 5 miles away from Yahoo!'s headquarters :))
[02:07]
gac410:) [02:07]
Babarlet me guess... IPv6
indeed. No clue why my laptop thinks it has v6 connectivity
[02:09]
gac410dns server? Resolving an AAAA record? [02:11]
FoswikiBotserver? Resolving an AAAA record? is . [02:11]
gac410That was helpful FoswikiBot
post-commit.pl just failed ... /libexec/ld-elf.so.1: Shared object "libaprutil-1.so.4" not found, required by "svnlook"
Looks like svn client needs to be recompiled again? I had to compile it earlier today to enable surf
git-svn just went belly up too: git: 'svn' is not a git command. See 'git --help'.
[02:11]
Babarok
gac410: can you try compiling it again?
Babar doesn't know where you did it, so...
[02:19]
gac410okay. hopefully it's in my history :)
Compiling now. Had to make clean, as it didn't want to rebuild
[02:20]
BabarCan't load '/usr/local/lib/perl5/site_perl/5.16/mach/auto/SVN/_Core/_Core.so' for module SVN::_Core: Shared object "libaprutil-1.so.4" not found, required by "libsvn_client-1.so.0" at /usr/local/lib/perl5/5.16/mach/DynaLoader.pm line 190.
git-svn is fixed, but doesn't work :)
[02:23]
gac410man, whatamess
okay svn is fixed.
[02:23]
Babaryeah, so is git-svn
which is called git-subversion, and conflicts with git... which is awesome, as some packages, like p5-Git-Repository, depend on git
[02:26]
gac410I did a make -C /usr/ports/devel/subversion ( clean, showconfig, make, reinstall ) [02:27]
Babarok. Next time, we'll use pkg
in fact, I'll do it right away
[02:28]
gac410Have to make sure "serf" is enabled
The one installed by gmc was missing serf, so had no support for http
I found the make -C instructions via google.
[02:28]
Babaryeah... and now that pkg uses binary packages... they do not have all the features [02:30]
gac410another fail in git-upgrade.pl resize: not found Unable to get Terminal Size. The TIOCGWINSZ i [02:31]
Babaroh, Vicky is back? That's cool!
what?
resize?! but that's a terminal thing. Why are we running this in a cronjob?
[02:31]
gac410I have no idea. not my cronjob :) [02:32]
Babarand why are you getting this output and not me? :)
yeah... iirc, it's mine. :)
[02:32]
gac410Because you were offline, and I finally found the email and changed it to me. After you upgraded perl we were without git updates for several weeks. Finally tracked down the missing logs. [02:32]
Babarnice!
and yeah, I set it to me as ... it was noisy
I think I've found the issue
[02:33]
gac410Feel welcome to send the logs back to you again :) [02:37]
Babaryeah... I need a server again
or more, I need my server as a server, and not as a useless piece of art under the TV
[02:46]
gac410:) [02:46]
........................................ (idle for 3h18mn)
***ChanServ sets mode: +o CDot [06:04]
.................... (idle for 1h36mn)
ChanServ sets mode: +o MichaelDaum [07:40]
MichaelDaumgood morning [07:41]
ModAcOstGood morning [07:50]
........ (idle for 39mn)
MichaelDaumMichaelDaum doing an svn up trunk ... which gives me a break to make coffee
500 internal server error on trac.f.o
[08:29]
..... (idle for 21mn)
***ChanServ sets mode: +o pharvey [08:51]
GithubBot[foswiki] FoswikiBot pushed 3 new commits to master: http://git.io/gLCMJg
foswiki/master 8a6d4a4 CrawfordCurrie: Item12820: correct computations for correct row and draggable positioning; change span to div to support browsers that get confused by blocks in inline elemens...
foswiki/master 68be3fc CrawfordCurrie: Item12820: make sure we can move to the end of the editable rows...
foswiki/master 0182561 CrawfordCurrie: Item12820: make sure we can move to the end of the editable rows...
[09:00]
***GithubBot has left [09:00]
FoswikiBothttp://foswiki.org/Tasks/Item12820 [ Item12820: Javascript inconsistent operation for edit operations ] [09:00]
MichaelDaumMichaelDaum reading up on the summary of the discussion at http://foswiki.org/Development/MoveCodeRepositoryToGit [09:10]
.... (idle for 19mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/oX_ADw
foswiki/master 77c91c7 CrawfordCurrie: Item12814: disable edit on tables that have no macro...
[09:29]
***GithubBot has left [09:29]
FoswikiBothttp://foswiki.org/Tasks/Item12814 [ Item12814: All tables get edit controls if one table uses the EDITTABLE macro. ] [09:29]
.... (idle for 19mn)
JulianLevensMichaelDaum: have you finished reading up on the summary of the discussion at http://foswiki.org/Development/MoveCodeRepositoryToGit?
The more I read about sub-modules the more I feel uncomfortable
[09:48]
MichaelDaumyes. seems so. I have to trust Paul and George on this as I haven't played with git for a long time. [09:52]
JulianLevensProblem is both of them are not entirely positive about sub-modules. Crucially no-one is coming up with a clear recommendation along the lines of: use got sub-modules for Foswiki this way and it'll work well
Don't get me wrong I do trust them. George in particular is doing a lot of good work here. If as a result he works out the Right Way To Do It, then I'll follow his lead
[09:55]
MichaelDaumyep. thats how I see it as well. [09:57]
JulianLevensAs a Git (and SVN) numpty what else can I do anyway? [09:58]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/svK1vg
foswiki/master 3d1f027 CrawfordCurrie: Item12772: missing dependency...
[09:59]
***GithubBot has left [09:59]
FoswikiBothttp://foswiki.org/Tasks/Item12772 [ Item12772: EditRowPlugin depends on JSON.pm?! ] [09:59]
***ChanServ sets mode: +o Lynnwood [10:00]
MichaelDaumI am okay with either way as long as it worx.
the only thing for me to make sure is how to get the latest and greatest from all repos / sub-modules.
for now I use the monolithic repo foswiki/foswiki which basically gets me a svn/trunk structure.
problem is when we need to fix one and the same error in lots of extensions, e.g. use locale
and then how is this tracked
maybe thats just simple as can be. I am simply not sure.
[10:00]
JulianLevensy, that was one of my thoughts. I was wondering about using git monolithic as a 1st step, with further refinements later
But I'll leave that to the experts
[10:04]
MichaelDaumdoing it right from the beginning is king.
doing it wrong having to stick to it is worse on a negative scale.
from what I see we are heading towards pseudo-install and BuildContrib taking care of the details underneath anyway
pseudo-install will create the repos or submodules as required ... as far as I know it already does so today
[10:06]
JulianLevensI don't think monolithic is wrong, we know that works, don't we? [10:09]
MichaelDaumnot sure which bits are missing for BuildContrib other than rev-hash mapping n stuff [10:09]
jastmonolithic works with SVN. it does *not* work with git. [10:10]
MichaelDaumit would be wrong if you had to revamp all of your checkout area just to split up a monolithic setup into a bunch of repos per subdir
jast, ah ok. why?
[10:10]
jastin svn you can checkout/merge individual subdirs. that's not possible in git
so, in git, when you do a 'git merge', it treat all changes you've done as part of the merge
so you can't merge changes in one extension without also merging the changes in all other extensions and core
[10:11]
MichaelDaumgood point. totally forgot about that. [10:12]
jastthere is absolutely no way around this in git
I know submodules are a hassle, especially if you have lots of 'em, but at least they _work_...
[10:13]
MichaelDaumboth submodules and one-git-per-extension solve the situation dont they? [10:13]
jastsubmodules *are* basically one-git-per-extension :)
they're something you can do in addition
a meta-repository that refers to all the individual ones, with the nice property that you can refer to specific versions of each extension
that way you can bundle the right versions together for a release, for instance
[10:14]
MichaelDaumsounds like getting lost in managing meta data [10:15]
jastwould have to be automated, I suppose [10:15]
MichaelDaummaybe just something new to learn and get used to [10:15]
jastthey're pretty much like svn:external [10:16]
MichaelDaumwew, cool feature didnt know it existed: ONLYIF not(eval('require Foswiki::Tables::Table')) [10:16]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Tables::Table [10:16]
jastfun [10:17]
JulianLevensjast: with something like the JQuery stuff would you propose a sub-module for all the JQuery stuff or a sub-module for JQuery and sub-sub-module for the JQuery extensions? [10:17]
jastgood question
is there a plan to split up JQueryPlugin further?
[10:18]
JulianLevensnot my baby, I'll pass that on to MichaelDaum [10:19]
MichaelDaumthere is cruft in JQueryPlugin indeed
things that should have been captured in a contrib if their own.
[10:20]
jastwell, anything that does get factored out, I'd just treat like any other extension [10:20]
MichaelDaummost of it came in as part oJQueryPlugin being added to the core as is
jast, right.
I don't think that it plays any role in structuring the git sub modules
unless there is a feature request to move out some stuff in extensions of their own...the normal drill.
only then will they get a sub module of their own
analog to the current dirs under svn/trunk
[10:21]
jastmakes sense [10:22]
JulianLevensJulianLevens Searching for git training material [10:24]
jastthere's the official book at http://git-scm.com/book
well, "official" in the sense that it exists and can be read online for free :)
[10:25]
JulianLevensy, I have a real printed copy too [10:26]
jastalso I can tell you that the official channel (#git) is quite helpful
(I happen to be a channel manager :))
[10:26]
JulianLevensFor git, something has really clicked yet
The basic workflow is easy enough
Whenever I need to move away from that or I get an unexpected message, I do struggle
[10:28]
jastthe channel is probably going to be fairly useful for that kind of thing
it's hard to find the right resources, but there are plenty of people there who know where they are :)
[10:30]
JulianLevensThanks, I'll remember #git when I'm next in trouble [10:31]
jastthis one may be helpful for understanding the way git is structured: http://gitolite.com/gcs.html [10:32]
MichaelDaumhttp://gitref.org/ [10:37]
............................ (idle for 2h17mn)
***gregg4567 has left [12:54]
...... (idle for 29mn)
ChanServ sets mode: +o gac410 [13:23]
gac410Hi Babar ... I'm seeing the resize errors on the git-svn-split.pl script looks like the updates are still making it to github though
And getting git warning: push.default is unset; its implicit value is changing, on git-upgrade.pl That one I can probably fix. ;)
[13:26]
...... (idle for 25mn)
MichaelDaumtrac.f.o is broken [13:52]
gac410hi micha. hm... I thought we checked that last night. [13:53]
MichaelDaumdunno just see a 500 internal server error ... could be anything [13:53]
gac410I guess not ;) Babar finished off the upgrade that gmc started.
Let me take a look - I helped some last night too,
darn... python upgraded to 2.7, and config got lost I think.
well not lost, left behind in old python libs.
[13:53]
MichaelDaumthe viper bites back [13:56]
gac410maybe not. .... hm.. [13:56]
MichaelDaum: I'm not all that well versed with python and wsig. Getting a load error.
ImportError: /usr/local/lib/python2.7/lib-dynload/_io.so: Undefined symbol "_PyInt_AsInt"
[14:01]
MichaelDaumseems to be more related to shared libs being incompatible ... a distro error I'd say in linux land [14:01]
gac410Not sure if it's apache mod_wsig needs update, or whatover...
freebsd changed from compiled to binary packages ... It's been causing some grief,.
[14:02]
jastsounds like that lib is outdated or needs to be rebuilt
I always stay far away from Python, though, so I can't contribute anything constructive...
[14:02]
gac410I had to compile svn yesterday to get http support added back in. The binary version didn't include http: :P [14:03]
jastlovely
to be fair, IIRC http is lots slower in svn than the native proto
[14:03]
gac410Luckily I found a google hit saying exactly what to do to recompile. no such luck with the python error.
well, we use http:// internally to check out trunk.foswiki.org so trunk was not being updated.
[14:04]
MichaelDaummaybe time to switch os [14:06]
gac410Windows yay
:D
[14:06]
MichaelDaumor distro or whatever to avoid having to compile software by hand on a production server [14:07]
gac410looks like mod_wsgi was built on 3/19 ... hm... but python is dated 3/12, so that's not it.
well as a gentoo user, I can't say compiled is bad. Just a bit time consuming :)
[14:08]
MichaelDaumdoesnt help to do a job quickly [14:10]
gac410no. though as machines get more powerful it gets faster.
anyway, for this particular situation, I'm just not all that familiar with python ... That's one language I try to stay away from.
looks like it might be internal to python. File "/usr/local/lib/python2.7/io.py" is what is failing.
[14:14]
Tempted to try to reload httpd ... but if it fails, I'll need help for sure.
translate.f.o and trac.f.o are now failing with the same error. Babar fixed the settings issue, but then hit the _io issue. But give we are about to switch to a new branch I told him translate wasn't all that urgent,
ping gmc are you around?
[14:29]
jasthave you gone through this already? http://code.google.com/p/modwsgi/wiki/CheckingYourInstallation#Python_Shared_Library [14:35]
gac410Some of it, yes. ldd resolves everything.
I didn't want to start changing around scripts within apache to do some of the runtime tests.
[14:35]
jastunderstandable [14:38]
gac410Apache sets an LD_LIB path... I just set that in my local shell and re-tried the ldd - still resolves everything,.
I don't get the same error though if I explicitly run io.py from local shell. So it is probably something in the apache env, not the python build
[14:39]
MichaelDaumto all: please test the new editor on trunk, i.e. the TinyMCE + NatEdit integration.
there seems to be a transient error where edits are lost when editing via tinymce.
as far as I've investigated the javascript code there might be an error in the tinymce-foswiki js part of TinyMCEPlugin
my understanding of this code still is not deep enough to see what's going on
besides not being able to find out about the circumstances which trigger the bug
it seems to be a race condition of some kind. at least thats how the error manifests.
[14:43]
gac410could something be getting cached ... trunk uses the cache by default [14:47]
MichaelDaumno caching involved
please try to install a local foswiki-1.2.0 alpha
I am filing an urgent bug against TinyMCE + NatEdit
blocking the release
Item12822
[14:48]
FoswikiBothttp://foswiki.org/Tasks/Item12822 [ Item12822: edit changes lost sometimes ] [14:51]
gac410That you've seen it on 1.1.9 is good to know. Trunk has a much newer version of tmce enabled, plus some patches to make it easier to try new versions. [14:53]
MichaelDaumI am investigating foswiki_src.js and foswiki_tiny_src.js to see whats going on
this code is awful and desperately needs to be rewritten from scratch
[14:57]
jastrewriting from scratch tends to introduce other subtle bugs ;) [14:58]
MichaelDaumalas I can't see the ramifications of some wild code in there, e.g. the crude callback mechanism in there [14:58]
gac410Hm... on trunk configure, is the "Your Wiki" supposed to take you to your webhome?
tmce has changed a lot over time. It could be that stuff got left behind as things changed.
MichaelDaum: Do you have any hints as to what triggers the save issue? I do very little editing on my local wikis, it's just a test machine.
I'll test, but it's generally contrived testing,
[15:02]
jastgac410: re "your wiki": yes [15:07]
gac410On my checkout it seems to just be a <nop> [15:07]
MichaelDaumgac410, I've added a few observations [15:07]
jastindeed, the button wasn't implemented completely. blame me. [15:09]
gac410:_
:)
MichaelDaum: re preview not showing changes. ... That might be the reason preview is disabled for wysiwyg on 1.1.9 pattern skin wysiwyg
That was done a loooong time ago though.
[15:10]
MichaelDaumworks fine normally. disabling it only works around symptoms. [15:12]
gac410Original preview bug was Item5263 Fix was to remove the feature. [15:16]
FoswikiBothttp://foswiki.org/Tasks/Item5263 [ Item5263: Preview with pickaxe/wysiwyg and Firefox make you lose all you typed when you go back ] [15:16]
gac410Which has a follow-on task Item10065 And brainstorming Foswiki:Development.EditorAPI [15:24]
FoswikiBothttp://foswiki.org/Tasks/Item10065 [ Item10065: Add preview for WYSIWYG ] http://foswiki.org/Development.EditorAPI [ EditorAPI ] [15:24]
CDotMichaelDaum: the "crude" parts are many releases of TMCE old. That's the way you (used to) have to do things. There's quite a lot of code in there I don't recognise (not touched it for years), but what I do recognise hasn't kept pace with TMCE releases. [15:25]
gac410And the 4.x tmce makes more major changes in how it handles hooks and callbacks. [15:26]
CDotmoving target :-( [15:26]
gac410yeah. I started an effort to version some of that code to track tmce. Would be easier to update new / test releases. [15:27]
foswiki_irc9New user here. How would one go about creating a new template page? [15:28]
gac410hi foswiki_irc9 what do you mean by template? A "starter" layout for a new topic in the editor?
unfortunately foswiki uses the word template to mean different things in different places. :( it gets confusing,
[15:29]
foswiki_irc9Correct. Just looking to create a new topic that can be selected from the drop-down so our team can input information based on a set of predetermined sub-headings [15:30]
gac410Hm... I think if you just edit a topic in that web, and name it ending in the word Template, it should show up on that list. [15:33]
jastyeah, that's the way it's supposed to work
and put it in the same web where the current templates are
[15:34]
foswiki_irc9I like simple solutions. Thank you. [15:34]
MichaelDaumbesides loading the wysiwyg editor in two steps is pretty slow [15:35]
jast(you can customize for each web where it looks for *Template topics, but as long as you don't need that it's probably more sensible to stick with defaults) [15:35]
gac410MichaelDaum: was that for me? I don't think I changed the load order. [15:37]
MichaelDaumno thats a more common observation while digging into TinyMCEPlugin [15:38]
gac410oh...
Just don't let the tmcevirus hit you. mtempest worked on it, and seems to have left the project. pharvey worked on it and has become scarce. Hope it's not a trend :)
[15:38]
MichaelDaumonce I know which parts are woth keeping, I can come up with a more maintainable version [15:42]
gac410One request ... keep it easy to switch out tmce versions. It's not too onerous currently, except for when tmce makes api changes ... which seems like far too often. [15:43]
MichaelDaumthe slow-by-design remark was about the content to be edited being transfered to the browser twice actually: once as tml with the original edit page, then again using yet another rest call ... which gives you this "please wait" message within the edit area.
this basically uses the same calls already used to switch back and forth between wysiwyg and wikiwyg
[15:44]
jastunfortunately cleaning up CKEditor for use with foswiki isn't any easier... CKEditor mangles the HTML in different ways, so WysP would need to be essentially rewritten to support quirks in multiple editors [15:44]
gac410That seems to be what pharvey started in on with the EditorAPI brainstorming.
gac410 wonders if it would be better to let a js component convert tml -> html -> tml customized for the particular editor chosen rather than doing it in the plugin ... if that's even possible.
[15:45]
jastone of our guys started work on that
some of the intricacies in TML rendering resulted in that project getting set aside, presumably forever
[15:47]
gac410:(
Unfortunately that round trip is horribly complex. CDot seems to be the only one who really understands it. I look away for 5 minutes and it's like I had a lobotomy.
[15:47]
jastthe only thing that is horribly complex is the logic in WysP that presumably attempts to preserve various details in TML :)
the really tricky areas in there are whitespace and entities/high-bit chars
[15:50]
gac410yeah... we still have serious whitespace issues in places. And %TABLE% macro corruption too. (well extra white space separates the table from it's definition at times)
I keep finding wordsrun togetherat times
But I gave up trying to fix. Every time I touch something 5 other things break. Last time I actually made some fixes, I think it took a half dozen wysiwyg updates to fix most of them.
[15:51]
jastthe really ugly stuff happens inside tables [15:54]
MichaelDaumthere are so many errors in tinymce itself related to deleting elements like headings [15:55]
jast(ugly in terms of conversion bugs)
there are so many errors in WYSIWYG support in various browsers
[15:55]
MichaelDaumdeleting a h1 at the top will leave behind a heading 1 paragraph marker
deleting a line before a h2 will convert the h2 into a span element being styled as a h2 look-a-like
[15:55]
gac410There is a lot of code to allow you to reach a line before the first table, or the line after the last table. ... and I still find I have to edit wikitext to reach them. [15:56]
MichaelDaumthere's no way to insert a line behind a verbatim block. clicking into that area won't set the cursor.
etc etc etc
[15:56]
gac410I'm not sure how much of that is us and how much is tmce. [15:57]
MichaelDaumtopics with lots of inline styles, even using <font> [15:57]
jastand how much of it is the browser [15:57]
MichaelDaumme neither. but we know how much this sux.
what we can do is to integrate tinymce or cke or whatever editor as good as possible doing our job on the foswiki fringe to it. we can't fix the editors' bugs for sure.
MichaelDaum currently trying to delete a h1 ... without success
[15:58]
gac410trunk and 1.1.x use different tmce releases. You can also switch trunk to a different TMCE using a Set ... [16:00]
MichaelDaumgawd the more I use the editor the worse it gets
try this: copy-paste the text within a verbatim area to a non verbatim area. what do you expect is the result?
[16:01]
gac410iirc, it creates another verbatim.
heck copy/paste from a different window on the system, shell, whatever, often creates a verbatim.
[16:02]
MichaelDaumtrying to copy a verbatim block is broken anyway. it generates a pre and inserts tons of brs [16:03]
gac410See http://trunk.foswiki.org/Sandbox/TinyMCEVersion4Edit for the sets to switch tmce versions. Though tmce 4.x is not functional. That will all have to be reverted once we branch 1.2 [16:03]
MichaelDaumwywiwyg dialogs are all empty
e.g. not possible to insert a link or image on trunk.f.o
[16:04]
gac410That's not good. I wonder if it's the tmce version change. [16:06]
MichaelDaum: I think this topic: http://trunk.foswiki.org/Sandbox/TinyMCE349Edit should be using the 3.4.9 editor that 1.1.9 uses.
Might help to see if issue is tmce release dependent.
3.5.10 is still the latest 3.x on tmce. 4.0.20 is current on 4.x series. trunk has 4.0.11 (not functional)
[16:12]
MichaelDaumokay switchign to 3.4.9
lots of 404 not finding the mce_popup.js and stuff
https://.../pub/System/TinyMCEPlugin/tinymce/jscripts/tiny_mce/tiny_mce_popup.js -> 404
has to be https://.../pub/System/TinyMCEPlugin/tinymce-3.4.9/tinymce/jscripts/tiny_mce/tiny_mce_popup.js ... where's that configured ?
[16:14]
gac410hm... Obviously I missed one. Actually if that is failing, then is 3.5.10 broken too? Maybe that's one of the issues.
There are some macros that I changed to render the version .. That was a long time ago :(
gac410 takes the bullet. That's still hardcoded in the pub/System/TinyMCEPlugin/plugins/foswikibuttons/colours.htm
ah.... I think I remember. If the code is found in the editor, then that specific version is used.
Yeah. that's it. /var/www/foswiki/trunk/TinyMCEPlugin/pub/System/TinyMCEPlugin/tinymce-4.0.11/plugins/foswikibuttons/colours.htm needs an equivalent in the 3.5.10 and 3.4.9 directories. Or we need to keep a tmce directory around for the "common across versions" versions.
gac410 made a hash of this for sure.
my rev 17114 broke it all.
[16:27]
MichaelDaumit is because the plugins still assume the old dir structure using relative links
so when they try to load mce stuff it fails
[16:37]
gac410Rev 17114 reverted a change, so that we only had to duplicate stuff if the tmce version required a different version. But I forgot that tinymce directory went away. [16:38]
MichaelDaumit still is there sort of [16:38]
gac410yeah. And my testing worked, because the old tinymce directory was hanging around when I check out trunk. yech.
CDot was pretty sure the attempt to support multiple tmce versions was misguided. Maybe it should all get reverted.
[16:38]
jastI would tend to agree with that assessment, without even using tmce myself
I've heard enough about tmce, though :)
[16:40]
MichaelDaumproblem is: we _need_ to get it working for the release [16:41]
gac410revert 17114 ... that should at least get the 404's fixed for now. [16:41]
MichaelDaummeanwhile I've spent another hour with edit+save trying to repro the error a client reported losing edit changes... [16:42]
gac410MichaelDaum: I'm working on reverting 17114 for now. [16:49]
MichaelDaumokay. let's see then. [16:49]
gac410wtf... my git revert just evaporated.
okay... committed. And a quick tmce test still worked.
If you still have trouble, revert Rev 17111 takes us back to before I touched anything related to multiple tmce versions.
[16:53]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/tE7jNA
foswiki/master c34830c GeorgeClark: Revert "Item12665: Revert some of multi-tmce changes"...
[17:00]
***GithubBot has left [17:00]
FoswikiBothttp://foswiki.org/Tasks/Item12665 [ Item12665: Support multiple TinyMCE versions, begin testing TMCE 4.x integration ] [17:00]
MichaelDaumsvn up running on two machines ... time for a break [17:01]
pseudo install fails [17:07]
gac410really? Worked here.
Hmmm failed on foswiki.org File::Copy::Recursive required dpenedncy. but that's not my change.
okay... updated cpan. trunk.foswiki.org just updated okay
I don't see any 404's in the net log on trunk.
[17:11]
MichaelDaumWARNING: Cannot find source file for ...foswiki/trunk/TinyMCEPlugin/pu/System/TinyMCEPlugin/tinymce-4.0.11/tinymce/js/tinymce/plugins/emoticons/img/smiley-embarassed.gif
see pu/System
[17:18]
gac410Oh... 4.0.11 is definitely not usable. I'm surpised it even loads. [17:18]
MichaelDaumerror in its manifest [17:19]
gac410oh... okay.
pseudo-install should still work, that's just a warning.
[17:19]
MichaelDaumthere are still http and js errors clickin on insert image [17:20]
gac410okay. I'll revert the rest. Then we are back to before I touched anything.
okay ... dcommit running now.
This is a big one. Removing multiple TMCE releases, dropping back to an old one.
tbh though, we state outright that running on trunk is not advised.
man svn is sloooooo
[17:21]
Oh crap.... CDot's changes to the commit check blocked some of the revert We now have a complete mess.
I have no idea if this actually went through. It blocked all the topics due to bad topicinfo.
[17:34]
CDotif it blocks one thing, it blocks them all. There can be only one etc. [17:40]
gac410okay, trying revert again. Takes about 10 minutes to get to the "check" ... If it still fails, someon with svn will need to revert. [17:43]
OH FFS... What is wrong with %META:TOPICINFO{author="ProjectContributor" date="1395682500" format="1.1" version="1"}%
Fails "version must be 1
It also fails with Use of uninitialized value in string eq at pre-commit.pl line 233.
CDot... can you revert rev 17111. I can't get past your exit and micha's production trunk is unstable.
gac410 gives up.... wanders back to work on taxes .. that's less painful. And then back to home automation.
[17:52]
CDotgac410: the easiest thing is to temporarily disable the check altogether. Reverting it is not the right way to go. I have already checked in several topics running through the topic check, so if there's an error it's not a general one.
I'm not aware of any way to tell svn that a checking is a revert
[17:56]
gac410I ammended my commit to fix the topicinfo and it still fails. Revert from git-svn really sin't a revert [17:57]
CDotcan yuo show me the first 2 lines of the failing topic? [17:58]
gac410%META:TOPICINFO{author="ProjectContributor" date="1395682500" format="1.1" version="1"}%
%META:TOPICPARENT{name="Plugins"}%
[17:58]
CDotand what's it failing on? What does the report say? [17:59]
gac410It was failing author, date and version.... I fixed all three, (was version 2) [17:59]
CDotand after you fixed it> [17:59]
gac410trunk/TinyMCEPlugin/data/System/TinyMCEPlugin.txt meta-data is incorrect; cannot check in:
version must be 1
it IS version 1.
[17:59]
CDotso I see
CDot runs the script locally in debug mode
[17:59]
gac410It also complained about all the attachments, I seem to have fixed them acceptablyt
But in truth I'm not really reverting, because the svn exit is making me fix the reverted topic.
But that's a git-svn artifact I guess.
Anyway, someone just revert the damn commit, micha will be happy and I can go away please...
[18:00]
CDotok, i ran the check locally on the file you gave me and it does not fail. So you missed something..... [18:03]
gac410I cut/pasted those lines from the commit. [18:03]
CDotfor f in <files being checked in>; do perl pre-commit.pl test -m "Item1234: test" $f; done [18:04]
FoswikiBothttp://foswiki.org/Tasks/Item1234 [ Item1234: MathModePlugin's latex2img not executable ] [18:04]
gac410gac410 is frustrated. he didn't know trunk was supposed to be used by production sites. Of course it's unstable.
I have an ancient SVN checkout. Let me svn up that one and revert from there.
[18:06]
CDotwhat checkin are you trying to revert? [18:06]
gac41017111 Dumping all my tinymce work [18:07]
CDotok [18:07]
gac410It's huge and git-svn takes 10 minutes or so to get to your check. [18:07]
CDotjust in TinyMCEPlugin? [18:07]
gac410It might touch build contrib ... Not certain. [18:08]
CDotCDot has to svn update before he can test the revert [18:08]
gac410my svn up is failing with all sorts of conflicts. [18:09]
CDotmine went cleanly. [18:09]
gac410My hadn't been updated in years :) [18:10]
CDotCDot is running the revert [18:11]
gac410my svn update finished with all sorts of tree conflicts. probably ought to just toss it. [18:12]
CDotsounds like it. Mine went clean.
CDot is running the pre-commit check locally on the reverted tree
[18:14]
gac410It initally found 4 issues. The 3 attachments don't have a user or date. And the topic didn't have right author, date or versino.
TBH I disagree with checking attachments for "current" date. It should be the date of the attached file.
[18:15]
gmcgac410: i'm around, what's up? [18:18]
gac410trac.foswiki.org is down due to python issues. missing symbol loading _io.py
both trac and pootle have same issue now. Babar fixed original pootle problem, it's config was in the python 2.5 lib tree
He aslo finished the update process
Oh... and I had to recompile subversion, it was missing serf option - no http support
[18:19]
CDotok, all checks passed locally (only had to fix TinyMCEPlugin.txt) [18:21]
gac410gac410 doesn't understand why my change didn't work. [18:21]
GuilainChi everyone, SvenDowideit, if you're around, first thank again for the debian repository on fosiki, but am I the only person who need to chown www-data:www-data data/* when using debian package (your repository) ? and if not, is not possible to include in install script ? [18:23]
gac410Did your svn revert make it into the repo? I did a svn rebase, didn't see any changes [18:23]
CDotnot yet. patience.
I need an item number.
[18:23]
GuilainCGuilainC wish good luck to solve the problem on the server... [18:24]
gac410Same item? Item12665 [18:24]
FoswikiBothttp://foswiki.org/Tasks/Item12665 [ Item12665: Support multiple TinyMCE versions, begin testing TMCE 4.x integration ] [18:24]
CDotCDot is committing
ok done
that didn't hurt.
[18:26]
gac410revert is the only area with git-svn that I've had trouble with, since it really doesn't revert. [18:27]
CDotCDot has gone to cook the curry that he was too late to start cooking last night :-) [18:27]
gac410have fun. [18:27]
CDot:-) [18:28]
gac410what time are you serving? :D [18:28]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/cU6tUQ
foswiki/master 555a542 CrawfordCurrie: Item12665: reverted George's changes, at his request...
[18:29]
***GithubBot has left [18:29]
gmcgac410: hmm ok, trac was ok when i left it :) [18:32]
gac410Yes indeed. I think Babar broke that one [18:32]
gmcso what exactly has babar been up to? just so i don't do things that undo his work.. [18:33]
gac410MichaelDaum_: Okay... trunk is back on TinyMCE 3.5.7 None of my changes remain. [18:33]
gmcghaa those choppers are driving me crazy [18:33]
jastoh hey, more activity
jast is at home now
[18:33]
CDotgmc: police? [18:34]
gac410gmc. He ran the rest of the updates, using pkgng I believe.
choppers - false teeth?
[18:34]
gmcCDot: some nuclear waste summit going on a few km from here [18:34]
jastobstacles you have to jump through to get to the end of the level? [18:34]
CDotI had one hovering over the house a few days ago. Some numpty ran from the police, and fell in the river. Took a fire engine, two ambulances and a helicopter to get him out. [18:35]
jastI discovered yesterday that it feels rather weird to walk right past a lineup of ~30 police officers, armed and armoured to the teeth [18:36]
CDotdid you pick out the culprit? [18:36]
jastit wasn't that kind of lineup [18:37]
CDotoooooh.... firing squad. yeah, those are nasty. [18:37]
jastthere was a salafist protest in the city I was visiting
about 100 protesters, and about three times that in police folks
[18:37]
CDoty, they've all learned the art of "kettling" now. Protest marches just ain't what they used to be. [18:38]
jastit wasn't even a march. they were just standing there... [18:39]
CDotanarchists! [18:40]
gmchippies! [18:40]
CDotreligious extremists! [18:41]
jasta few white supremacists were present, too [18:42]
gac410Okay... for anyone who cares, the TinyMCEv4 work I did is out on https://github.com/gac410/foswiki/tree/TinyMCEv4 [18:53]
GuilainCwhat policy on task, when you find a a bug needed a task similar to an other, but which not apply on the same code base : use the already task or make a new one
for details see : http://foswiki.org/Sandbox/TwistyTest related to http://foswiki.org/Tasks/Item10986
[18:57]
FoswikiBot[ Item10986: Use of indented TWISTY breaks page formating and places LeftBar on right side of screen. ] [18:58]
GuilainCthanks bot [18:58]
gac410that task was closed ... [18:59]
GuilainCNo Action Required is closed for yo
you
so, for you, no need to change anything
all right ?
[19:01]
gac410Yes, The issue is how lists and twisties are rendered. Some constructs can't be supported.
Can you put the twisty complete on one bullet line?
HTML has restrictions on what elements can be inside other elements,
[19:02]
gmcok, i broke the webserver [19:03]
gac410Congratulations! [19:04]
GuilainCcongrat's gmc
its bsd isn't it ? it made for it :)
thanks gac410, so IMHO, it's need to document it into the doc ?
isn't it ?
[19:05]
gmcaaand it's back..
GuilainC: well, i've seen people get debian in this state as well .. :)
[19:06]
gac410Yes I agree. GuilainC [19:06]
GuilainChow could i purpose an documentation uprade ?
uprade = upgrade
he gmc, i know well, passing to the last version of debian, make me nightmare
[19:07]
gac410In that case, it is fine to add a note to the task asking for it to be documented, and re-open the task. If you want to, add a comment and I'll open it. [19:07]
GuilainCfirst time in the last of 15 years
it will be a pleasure to help, where i can, documentation ;)
[19:07]
gac410I like the recent gentoo changes. When you build a library, it auto preserves the old version, and tells you to build @preserved_rebuild and when all dep's have been built, only then does it remove the preserved libs. [19:21]
..... (idle for 20mn)
GuilainCtaking time to see http://foswiki.org/Tasks/CurrentState, and observe that it very source code oriented, and there isn't any state for documentation enhancement, i'm right ? [19:41]
gac410Just set it to "Confirmed" That should be fine.
The docs are a bit of a twilight-zone. We really ought to allow doc updates via the wiki. We are a wiki after all.
But we don't have a mechanism to feed them back into subversion.
That's an area where github will probably make things easier.
[19:42]
GuilainCyou ahead of my other interrogation (why doc are not made by wiki)
and you need subversion/git for packaging ?!
should i write documentation -on this point- into subversion/git too ?
[19:47]
gac410We use Subversion so we can go back to the complete state of everything - docs, code, etc. at any point in time.
If you are interested in doing foswiki development - even docs, and want to go to the effort of getting a subversion checkout, yes we can give you authority to check in doc and other changes.
What we really need is a way to enable wiki updates of System. and a way to pull them back into a svn revision. That would make it much much easier to contribute.
[19:48]
GuilainCunderstand, using all the day git, is it possible to wait the git migration ? [19:50]
gac410I hope so... Who knows when it will really happen :D The proposal was made 3 years ago [19:51]
GuilainCunderstand well, the need, on server, it's not seems to be difficult
:) any date in mind ? even not the real one ?
for making checkout, commit & so, i should be able to do that
[19:51]
gac410We control the System web with subversion. And the "default" topics in Sandbox, Main, [19:51]
GuilainCon my project(s) what i do, is auto-commit every x time
auto-commit is without user, as often changes is produced
[19:52]
gac410We've been making good progress. I think we still have decisions to make on git structure. [19:53]
GuilainCcould we imagine a kind of similar functioning ? parsing, the System/directory, and auto-commit to svn ? [19:53]
gac410One issue with commit from result of wiki edits, we currently require that topics we ship always be version "1", by author ProjectContributor. [19:53]
GuilainCoki, for git structure, what the big change for you ? [19:54]
gac410I'd rather that we just accept the topics as edited, and fix them up when building a release.
monolithic repo (all extensions + core) or hybrid (monolithic core + default ext) or Separate repo for each extension including default extensions.
[19:54]
GuilainCyes i've found this version 1 by ProjectContributor, is not possible for each version packaged, to erase history and force the revision to 1 and autor to ProjectContributor
make repo as small as possible :)
[19:55]
gac410The script already does that... so I'm not sure why Crawford made that change. [19:55]
GuilainCit's my 3 years git conclusion :) [19:55]
gac410Yeah. I think we are leaning to the hybrid. One repo for what we release in a Foswiki tar.gz package, and then each extension in a separate repo.
But that needs to be implemented yet on github.
[19:56]
GuilainCwhat should be more coherent with foswiki structure, is to have hybride (core + contrib) + extension separately
don't know the admin side of github
[19:57]
gac410Yes indeed. Another issue, we have 17,500 commits roughly, each tracked in Tasks web using a InterWiki link Foswikirev:12345 [19:57]
FoswikiBothttp://trac.foswiki.org/changeset/12345 [ Changeset 12345 – Foswiki ] [19:57]
gac410We need a way to map the history from svn rev r12345 to a github.com/user/repo/commit/xxxxxxx hash
I've got a plugin / mapper partly written... and then discovered that cross-branch commits exist in svn, which means one svn rev points to two separate git commits. one in each branch.
[19:58]
GuilainCyes sure, but i thinking, that some utilities, be able to import all history of svn into git, and when done, just have change our interWiki link ? isn't it ? [19:59]
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/_DgV2w
foswiki/master 8e2a25d FlorianSchlichting: Item12823: Add Auto-Submitted header to MailerContrib email templates...
foswiki/master d74c771 FlorianSchlichting: Item12823: Add Auto-Submitted header to registration / password-reset email templates...
[20:00]
***GithubBot has left [20:00]
GuilainCah sorry, i've just understand the r12345 and hash link [20:00]
FoswikiBothttp://foswiki.org/Tasks/Item12823 [ Item12823: Mark auto-generated emails according to RFC 3834 ] [20:00]
gac410Yes. That's true... We already do that with git-svn. The git logs include the svn rev [20:00]
GuilainCyes in git you've to think by sha1 hash
by branch
undertstand all of the work,
so for helping in documentation, it better that i make my first step in svn :) in order to done it not in a long time :)
[20:01]
gac410the other thing to know... we are mostly working in the trunk checkout now. Foswiki 1.1.10 is not really being worked on. [20:03]
GuilainCfor information, your link with trac.foswiki is really good, using git, we have thinking with other solutions, but nothing is validated ; integration foswiki and git is still in progress [20:03]
gac410We need to branch trunk into a Release01x02 branch. [20:03]
GuilainCi suppose that doc is share between branch ?!
so no problem
[20:04]
gac410No unfortunately not. [20:04]
GuilainCah
:(
ok, i will do copy/past ;)
old school
[20:04]
gac410Currently on svn, each major release is a separate tree. [20:05]
GuilainCok noted [20:05]
gac410We have trunk (the future 1.2.x / 2.0.x ) Release01x00 long since dead 1.0 Release01x01 currently 1.1.9
We tag each patch release. So Release01x01x09 is a tag in the Release01x01 branch.
[20:05]
GuilainCclear
so first, do change in trunk
what i've to remember
[20:06]
gac410I'm thinking with git, we would actually branch Release01x01 -> Release01x01x10 for example.
We have a catch-all task for documentation. Item9693 No sense opening a task for each doc change.
[20:07]
FoswikiBothttp://foswiki.org/Tasks/Item9693 [ Item9693: Documentation updates for Foswiki 1.2.0 ] [20:08]
gac410If a user reports a doc issue, we'll fix it on the reported task, like your update to the Twisty task. But when we spot typos, unclear stuff, we just use the generic doc task. [20:09]
GuilainClogic [20:12]
gac410CDot: When you are back and done with your curry... I think the decision to scrub TOPICINFO on checkin is not a good one. [20:13]
GuilainCon git, i don't really know how merge between branch is easy or not in svn, but in git, i use everyday the branch merge
using, three main branch development : dev, testing, and stable
[20:13]
gac410I've not used svn in a long time, I'm mostly on git-svn for foswiki. cherry-pick is wonderful. [20:13]
GuilainCdon't forget that in git, a release is normally a tag as "v1.1.10"
cherry pick ?
[20:13]
gac410git cherry-pick <commit-hash> applies a commit from any other branch into the current branch including the commit message
How do you merge a commit from another branch?
I usually just cherry-pick all of the commits I want.
[20:14]
GuilainCi let my developer do it :)
and they do well
[20:15]
gac410Ah... :) [20:15]
GuilainCbug i'm using tortoise, and i personnaly do merge branching rarely
let me check
[20:16]
gmcso now obama and all the world leaders are having dinner less than 2km away from my home.. argh, even more choppers [20:17]
GuilainCi'm only using merge with "branch" on tortoise, but thanks gac410, you let me discover git cherry pick, i will ask my developers tomorrow :)
gmc the question is : have they better than your curry ? :)
[20:19]
gac410That's CDot making the curry [20:19]
GuilainCso, it's to go back home, perhaps see you in some time, or tomorrow
ah ok gac410
;)
[20:20]
gac410Okay... g'night GuilainC [20:20]
gmci had pizza today :) [20:20]
............... (idle for 1h11mn)
foswiki_irc8Hey guys , I started to install the Foswiki http://foswiki.org/Extensions/DebianPackage this tutorial and when I do the first command # Apt-get update returns me a problem: "W: GPG error: http://fosiki.com stable Release: The Following Signatures Could Not be verified because the public key is not available: NO_PUBKEY 379393E0AAEE96F6" And can't understand it; Another question I'm following the correct tutorial because thi [21:31]
FoswikiBot[ fosiki: foswiki consulting, development and support ] [21:31]
gac410hi foswiki_irc8 Those packages are maintained by SvenDowideit ... Sven, any ideas? Are you around [21:39]
foswiki_irc8gac410 , i don't know, I installed again and managed to get to work, but thanks [21:41]
gac410Okay great. I think they are signed with his key, I use mine for the official tar/zip files, but he does the debian pkgs
You may need to chown on the data directory - I've seen other reports that it doesn't set the correct www-data ownership,
[21:42]
......... (idle for 40mn)
***gregg4567 has left [22:23]
............ (idle for 59mn)
ChanServ sets mode: +o pharvey [23:22]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)