#foswiki 2013-10-24,Thu

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

WhoWhatWhen
***gac410 has left [04:45]
.................. (idle for 1h25mn)
ChanServ sets mode: +o CDot [06:10]
.................. (idle for 1h29mn)
foswiki_irc3hi, what do I have to do, to create a topic with the 'save' CLI script with a formtemplate. Because when I add the parameter 'formtemplate' and want to fill some fields with the parameter <any name>, the topic is created without the form
or do I have to use the 'edit' script, but that doesnt seem to work as well
[07:39]
foswiki_irc4sry had a disconnect
did someone answer to my question regarding the save sciprt?
[07:41]
....................... (idle for 1h54mn)
kip3ffoswiki_irc4: if you are running this from CLI, then you want to be using the save script not edit
the edit script only makes sense as an HTTP request
could you use pastebin to document explicitly what commands you have tried, and what they did?
[09:36]
***ChanServ sets mode: +o Lynnwood [09:48]
foswiki_irc4ok sure. one second pls [09:49]
http://pastebin.com/d961YAPr here it is
the -Summary parameter is a field of the form EventTopic
[09:58]
kip3fso if you remove the -Summary argument, are you saying that the form is attached? [10:06]
foswiki_irc4no its never attached [10:08]
***ChanServ sets mode: +o MichaelDaum [10:09]
kip3fare you able to make a page with form attached from the web interface? [10:10]
foswiki_irc4jeah that works fine [10:10]
***ChanServ sets mode: +o MichaelDaum [10:11]
kip3fhmm, i'm not sure what's wrong. I would recommend to test using the templatetopic argument [10:11]
foswiki_irc4ok I will try that. Thanks [10:12]
kip3fhth [10:12]
.... (idle for 15mn)
MichaelDaumhttp://perlnews.org/2013/10/twiki-6-0-0-released/
"user base consists of millions of users on tens of thousands of installations"
... counted via https://www.google.com/search?q=inurl:twiki/twikipreferences ... hm, that's 3220 as of today
[10:27]
foswiki_irc7does not work unfortunately @kip3f [10:32]
kip3fanother idea - what if you put the template topic in the same web
MichaelDaum - interesting way to look for *wiki installations. I'm surprised by how many TWiki sites are still out there
that haven't converted to FW
[10:33]
MichaelDaumthe press release is based on claims on wikipedia, which cites this page to substantiate it http://twiki.org/cgi-bin/view/Codev/NumberOfTWikiInstallations
side note. only the foswiki article is tagged "This article may contain improper references to self-published sources." ... the twiki one is not.
however, using the same method to count installs, foswiki scales even lower, to no surprise.
[10:37]
foswiki_irc7the strange thing is, that I cant see the form when I click on the topic but when I click on edit the form is attached. Does that mean I have to run the edit script after the save script. But I have not managed to make this work [10:41]
kip3fthere could be a skin setting that is preventing the form from being displayed during view
can you can pastebin the .txt file directly after you run the save script
foswiki_irc7 - sorry I have to go now
[10:43]
foswiki_irc7no problem. thanks for your help [10:46]
***kip3f has left [10:47]
....... (idle for 31mn)
MichaelDaumMichaelDaum reading TWiki::Mdrepo ... wtf [11:18]
....... (idle for 34mn)
pharveyMichaelDaum: http://twiki.org/cgi-bin/view/TWiki06x00/MetadataRepository [11:52]
.... (idle for 17mn)
MichaelDaumpharvey, yes seen that. yet still I dont get it. [12:09]
pharveyIIRC it's a giant system for organizing, managing webs. And speedy context-sensitive display of webs. And ownership/moderation of webs, etc. [12:10]
MichaelDaumcould be done as a plugin? [12:11]
CDotMichaelDaum: when using FCGI, does each instance have a fresh CGI object?
and is the instance created before, or after, the first Foswiki object is created?
[12:13]
MichaelDaumCDot, they are separate processes. [12:13]
CDotok. To control CGI, I have to know what process the CGI object is created in
reason is that subprocesses created when opening safe pipes are reaping parts of the CGI space when they terminate
[12:14]
MichaelDaumah the upload bug [12:15]
CDotspecifically, temp files (uploads) are destroyed if you call Sandbox::sysCommand
y, the upload bug. It's a bit of a bugger.
[12:15]
MichaelDaumhm [12:16]
CDotactually, I just realised, i know how to do it safely [12:17]
MichaelDaumI rewrote an uploader backend in TopicInteractionPlugin and came across something similar.
when uploading attachments in chunks
each chunk can be handled by a different (fcgi) backend... appendign a temp file
[12:17]
CDotsounds very similar [12:19]
MichaelDaumbut anyway. this impl is tightened to the specifics of plupload
it sends an incremented id for each part. these are cached on disk until the last part was received. then it assembles the thing and moves it into the Store
it gets each chunk from the $request->upload('file') stream and then does a sysread
... something otherwise done by the store
more precisly in Foswiki::Meta's attach()
CDot, but where's the sysCommand involved in that code path?
[12:21]
CDotwhen the user mapper loads WikiUsers, it uses readTopic
if you have selected RcsWrap, there will be a fork, and any multipart form in the CGI object will vbe blown away
(not the whole form, just the uploads)
I posted a trivial example to http://foswiki.org/Tasks/Item12084
[12:26]
MichaelDaumah okay made it thru the topic reading all of the issue ...
CGI.pm sux
[12:29]
CDotyup [12:30]
MichaelDaumanother way to see it is that there is a fork() before all posted data has been taken over from CGI [12:31]
CDotright
I tried to think of a way to "protect" the data, but I couldn't see any way without making a second copy of every upload :-(
[12:32]
MichaelDaumthats what I was musing about
take it out of the hands of CGI as soon as possible
there's some stream inside the request? wew, safe it for later.
^afe^ave
that's the way TIP does it
first collect all data, then hand it over to the store.
this let me implement the multipart uploader
but thats too much of a change for 1.1.9
[12:33]
CDotagreed [12:36]
MichaelDaumyour patch makes sense to me [12:36]
CDotgood! it took several hours to nail that one down :-( [12:37]
MichaelDaumany way to use method overloading instead of monkeypatching?
I mean Foswiki::Request ISA CGI
argh oic
[12:38]
CDotnope :-( [12:43]
MichaelDaumwell done solving this! [12:51]
CDotthanks! It was actually quite fun, in a murder-mystery kind of way.
you-know-who also has this bug in their codebase. Wonder how long it'll take them to pick up the fix.
[13:03]
..... (idle for 24mn)
***ChanServ sets mode: +o gac410 [13:31]
gac410Wohoo! CDot you found it [13:31]
CDoteventually yes. Just writing up a bug report for the CGI.pm maintainer. [13:31]
gac410Good catch. How on earth did you figure that out. Resort to the debugger? [13:33]
CDotElementary, my dear Watson. Careful detective work. [13:38]
gac410I knew something was triggering the close, but never would have guessed sandbox. I was figuring it was a try/catch somewhere that I was missing. [13:40]
CDotyes, you put me on the right track. Once I worked out how to reproduce it reliably, it was only a matter of time. No idea how to unit test for it, though :-(
The bug is, of course, in CGI.pm, so it's really their problem.
[13:41]
gac410I still don't understand how "damaging" the WikiUsers user list seemed to trigger the recreation for me anyway. [13:43]
CDotme neither, but it may just have been coincidence [13:44]
.... (idle for 18mn)
MichaelDaumdamaging in the sense of the store going on a slow path calling rcs for more rev info
touch WikiUsers.txt should suffice to trigger it
I gues
s
[14:02]
CDotgood point [14:05]
gac410Darn, you are absolutely right. Heck even the act of modifying the file to "damage" it triggered the rcs fork. So a very indirect path, but indeed that does it. [14:05]
CDotsomething to watch out for. WikiUsers really needs to stay off that path (as does everything, really) [14:06]
gac410As long as the caches are all loaded. Also it seems to only happen if using login names, where wikiusers is needed to translate from login name to wikiname. [14:08]
MichaelDaumthis ...
sub noCheckinPending {
+return 1 ;+
should have fixed it too
darn diff pluses
(on trunk)
[14:08]
............ (idle for 57mn)
CDotgac410: re: the apparent mismatch in trunk/Release01x01 versions - it looks to me like you made documentation checkins on Release01x01 that were not repeated in trunk. Everything else looks ok.
specifically http://trac.foswiki.org/changeset/16501
[15:06]
gac410CDot, it's actually a MailerContrib Config.spec that is mismatched, [15:07]
CDotThey're identical.... [15:09]
gac410My changeset http://trac.foswiki.org/changeset/15940 was partially reverted by http://trac.foswiki.org/changeset/16867
And I fixed it last night
Specifically http://trac.foswiki.org/changeset/16867/trunk/MailerContrib/lib/Foswiki/Contrib/MailerContrib/Config.spec
[15:09]
CDotyou are referring to the extra @@? [15:10]
gac410Well, sort of.. It's the trailing $ that breaks MailerContrib if the example is used
If you try that example (I even added a unit test) The regex compiles to a ...$+.... which croask
croaks
I removed the trailing $ and added the @@ to make sure it really couldn't match an email address.
[15:10]
CDotok, I guess I screwed up somewhere when I was trying to resolve the differences between the two branches. I recall having a tough time of it.
it all looks good to me now, though
[15:12]
gac410The original task http://foswiki.org/Tasks/Item11918 fixed the regex and did commit to both branches. Anyway, no big deal, and yes it all seems fine.
I synced up a bunch of differences of all core extensions last night, the remaining ones must NOT be synced up, as they are trunk-only.
At least they need to be very carefully reviewed,
Some are changes to the bin/ scripts, and how they find the paths, Babar's changes we decided to not use in the release branch.
[15:13]
CDotthere's a difference in the DEPENDENCIES, but it's an OK difference. [15:15]
gac410TablePlugin, TinyMCE, Wysiwyg, and a couple of others also have some major changes IIRC. I've been working on something else this am. My home automation stuff locked up :(
yeah.. I've been running this on each extension:
find WysiwygPlugin/ -type f -exec git --no-pager diff master:{} release11:{} \; | less
Most were indeed my screwups, In the heat of building 1.1.8, and maybe 1.1.7, I neglected to cross-sync my rev# / history cleanups
The easy ones were all synced up last night
TMCE on trunk uses a different version of TinyMCE. Actually I have an update to 4.0.8 stashed, that I'll apply to trunk once 1.1.9 is out.
We decided not to change TMCE in 1.1.8, and I don't want to risk a bit version jump in 1.1.9 either.
[15:15]
CDotok [15:20]
gac410EmptyPlugin has big changes, but that's API changes, so those can't go.
CommentPlugin was redesigned for trunk IIRC, big differences
PatternSkin ... big changes. We can't change the skin in patch releases
TopicUserMapping - bit changes in the User Registration pages, AC's work ...
er ... s/bit/big/
I think TablePlugin had big changes on trunk, didn't you start to refactor it into core?
SlideShowPlugin on trunk is major differences. IIRC it was redesigned, and Lavr rejected it for a patch release way back when. Needs to wait for 1.2
FamFamFam contrib has big changes, I think it's a new icon set, but also might be rendering changes, The change of all HTML on trunk from double to single quotes causes havoc
Yeah that reminds me, the " / ' change, that is all over the place on trunk, so we need to be careful.
TwistyPlugin *might* be safe to sync up. I don't understand the changes,
css and template changes,
Anyway cdot, I think I'm ready to start on 1.1.9, now that you've conquered the beast that is CGI
[15:22]
CDotexcellent; there's quite a queue waiting to go [15:32]
gac410Hm... It looks like an ancient task - http://foswiki.org/Tasks/Item9515 is missing from TwistyPlugin on trunk, So my suspicion is that AC removed some backwards compat css on trunk
So not missing, but has been reverted?
Specifically trunk does not have http://trac.foswiki.org/changeset/8983/branches/Release01x01/TwistyPlugin/pub/System/TwistyPlugin/jquery.twisty.uncompressed.css any more
So we can't sync that one up either. I suspect it's part of AC's grand reworking of the skin in 1.2
[15:35]
MichaelDaumjquery.twisty.uncompressed.* is the core of the plugin
ah no jquery.twisty.uncompressed.css is empty
there are two twisty implementations part of this plugin: one non-jquery, the other jquery
the non-jquery version is used when JQueryPlugin isn't enabled
we don't need the non-jquery version anymore
this plugin hasn't received enough love for quite some time
[15:39]
gac410The problem is that the plugin needs to continue to work on old releases, since we have no way to version the Extensions repo [15:46]
MichaelDaumright: foswikies without a JQueryPlugin ... which release was jQuery added to the core? [15:46]
gac410This will be the most difficult part of 1.2, All of the major skin changes, changes from double-to-single quotes, [15:47]
MichaelDaum... SlideshowPlugin [15:47]
gac410That one is a mess. Lavr was really upset about the new version, iirc [15:48]
MichaelDaumI am 100% with him on this one. [15:48]
gac410I don't use it so can't go by me [15:48]
MichaelDaumI've botched the current one in 1.1.x to look nice yet still backwards compatible. [15:48]
gac410ALL ... Does anyone have anything pending that should go into a BUG FIX ONLY! 1.1.9 ?
CDot just closed the last blocker, so we should be good to start the release process
[15:49]
MichaelDaumgac410, better email a deadline rather than asking here! [15:49]
CDotnice one [15:49]
gac410Well, My last email about the release meeting got exactly zero replies. Though a couple of people did contribute to the chat. [15:50]
MichaelDaum... they could have replied with a silly "I'll be there" ... which is umpf imho. instead they showed up ... which is yay. [15:51]
CDotif there are any pending bugfixes not reflected in the tasks DB, then those are the reporter's lookout. You can't be expected to last-minute chase. [15:51]
gac410CDot: yup ... I think I've used "due diligence" :) [15:52]
MichaelDaumtrue. still, I'd feel better with some timelines in general. even when they shift a lot.
in addition, emailing about the upcoming release increases awareness.
[15:53]
gac410Yes. How about I build 1.1.9 alpha and install it on foswiki.org, and attach it to the download topic. Email announcing that, with plan for first public beta a week later. [15:57]
MichaelDaumthat'd be great. [15:57]
gac410That way anyone who wants to try the alpha can download it, and in the meantime we get a bit of soak time.
I'll do one more gettext run, though I don't think strings have changed much at all.
[15:57]
MichaelDaumand people can start queuing up infront of foswiki shops [15:58]
gac410I think I've done all the syncing and version bumps on the extensions that are necessary.
There are seem to be a lot of backlevel fw's out there. Like the email exchange, and the site upgraded to 1.1.6, apparently ignoring the CVEs.
[15:58]
MichaelDaumdon't call it alpha. it is actually a release candidate.
alpha means: still api changes ahead.
beta means: api freeze yet features not totally bug-free
release candidates are the real thing giving people a last occasion to hammer it
as such patch releases never have alphas or betas.
[16:00]
gac410Hm. I view alpha as "test at your own risk" beta meaning "Please test, we think this looks good" and RC meaning, unless you report something, this WILL be the release,
I've followed the alpha / beta / RC cycle for every patch release I'v RM's
[16:02]
MichaelDaumalpha: we are not yet sure what we are doing while burning VC
beta: try this to get the feet wet
[16:03]
gac410To me an RC cannot change between RC and Release, and changes are not expected. I still expect translations. [16:04]
MichaelDaumboth - alpha and beta - are about features [16:04]
gac410:P How many angels on the head of this pin. RC means there is absolutely nothing in the pipe. This is the golden "candidate" that will go to press. I can't say that because we want the translators to work. [16:05]
MichaelDaumthey express the amount and degree of maturity of implemented features
ah there are translations pending?
[16:06]
gac410Hm.... I'll split the difference. I'll skip alpha. Beta 1 goes to Foswiki.org and available for download. RC will happen once translations complete, and no beta bug reports need fixing.
Yes
[16:08]
MichaelDaumoic [16:08]
gac410I was going to delay public beta until the 3 remaining strings are translated. [16:09]
MichaelDaumhttp://foswiki.org/Community/TranslationTaskTeam = AndreUlrich(*), WillNorris, SebastianKlus, CarlosBarros, OliverKrueger, ChYang, CoriaXu, TolgaYaman [16:09]
gac410I though german was complete, Now 5 strings remaining for german, 8 for most of the other "reasonably complete" translations,. [16:11]
CDothow do you tell? [16:11]
gac410http://translate.foswiki.org/projects/foswiki/ The summary shows 5 or 8 words remaining.
The ones with 100's or 1000's I've not been bothering with.
[16:13]
CDotCDot doesn't understand the UI. I can probably source translations for several languages, but I need to be able to ask the q simply.
looking at norwegian, for example....
[16:15]
gac410If you click the "8 words need attention" That will step you through the words. But unless you are an "authorized translator" all you can do is suggest [16:17]
CDotsure, but i don't see where to suggest
some of the words e.g. "disabled" can simply be googled
[16:20]
gac410Ah... Are you logged in? It has it's own account system. I turned off anonymous suggesting, as we were filled with spam [16:22]
CDotaha
any idea what my username might be? It rejects the usual
[16:23]
gac410Hang on I'll look [16:24]
CDotta [16:24]
gac410You are not listed, so never registered. [16:24]
CDotk [16:25]
gac410LarsEik is the only user authorized for Norwegian [16:26]
CDotLooking at it, he has made a bunch of suggestions that look correct to me. How do we accept them? [16:29]
gac410I think only he can "accept" them. I can add you to the project as a translator, and then you should be able to accept them. [16:29]
CDotok [16:29]
gac410Okay, You now have review authority for Norwegian
(might need to re-login - I'm not sure how it all works)
Once the translations are accepted, the 4-hour cron job should commit them back into SVN
[16:30]
CDotok, thanks [16:31]
gac410I think the last run was around 45 minutes ago, so need to wait till next "server time / 4" interval [16:32]
CDotstill can't see how to accept suggestions :-( [16:41]
gac410Don't you have a green "Submit" button to the right of the translation? [16:42]
CDotno [16:43]
gac410Hm... Let me look at another language, maybe I didn't give you enough permissions
Are you sure you are on the "Foswiki" project and not "Everything foswiki" which is bogus?
http://translate.foswiki.org/no/foswiki/core.po/translate/
Babar added "Everything Foswiki" to begin to handle translations for extensions, but it was never really implemented.
You have the exact same permissions as MichaelDaum who has been able to translate German just fine.
[16:43]
CDotah, you didn;t warn me about bogosity [16:47]
gac410I gave you the URL to step by the bogo-factor
:)
[16:48]
CDotI'm a geek! I have to know how *everything* works before I can do anything! [16:49]
gac410Note that the (Disabled) suggesion is missing the parenthesis, so they need to be added to the translation [16:49]
.... (idle for 18mn)
CDothmmm, I thought I had done that correctly
y, I had.
[17:07]
gac410Im back. Yes you did, I didn't check, I was just commenting on something that I probably would have missed.
Wohoo Norwegian is 100%
[17:09]
CDotmost of the languages are very very close - 2 mins work for a native speaker (or even a competent learner) [17:13]
gac410yes indeed.
Change in plans. It's so close, I'll send out an email to Foswiki-discuss asking for translation team to finish off the remaining.
[17:13]
JulianLevensCDot, SvenDowideit: I've updated http://foswiki.org/Development/StoresShouldBePassedConfigHash and would really appreciate more feedback [17:26]
gac410How does this sound: Private test build:  24 Oct 2013 (and install on foswiki.org) Release Candidate: 31 Oct 2013,  or when translations complete, whichever is first Release: 15 Nov 2013
[17:36]
.... (idle for 15mn)
CDotperfect! [17:51]
gac410email sent, bcc'd to the translation team members. [17:52]
..... (idle for 20mn)
I've updated the crontab to run the translation tools every hour. [18:12]
I changed the project name to "Everything in Foswiki (Not Used)" Maybe that will help in the confusion. [18:22]
......... (idle for 44mn)
MichaelDaum: are you still around? Just checking, my trial upgrade to 1.1.9 had a lot of jquery files removed from the distribution compared to 1.1.8 [19:06]
MichaelDaumhi gac410. I'm here.
which ones are these
[19:07]
gac410jquery.effects, superfish, A bunch of localization stuff: pub/System/JQueryPlugin/plugins/validate/localization/messages_ptbr.js [19:08]
MichaelDaumah yes [19:09]
gac41042 files overall all in jquery
wikiword shrinkurls ...
[19:09]
MichaelDaumerm? [19:10]
gac410pub/System/JQueryPlugin/plugins/wikiword/jquery.wikiword.init.js along with uncompressed and .gz variants [19:11]
MichaelDaumcould you be more precise or how could I check myself? [19:11]
gac410I use colas foswiki-upgrade-check with some tweaks - to find / remove files that are removed from the distribution. Hang on and I'll post it
The output that is
[19:11]
MichaelDaumupstream plugins changed a lot.
for instance jquery-ui now have a more consistent naming of their files
[19:12]
gac410http://pastebin.com/9WMVjhdW
So a questions for the release notes. Do "typical" upgraders need to remove those files or will everything be happy with the stale stuff in place?>
I use the upgrade script which always cleans out obsolete files.
[19:12]
MichaelDaumthese are all okay [19:13]
gac410Okay good. Okay to remove ... and okay to leave clutter too? [19:13]
MichaelDaumboth is fine. the former preferable of course.
the perl stubs take care of what is loaded
[19:14]
gac410Okay. Maybe the release notes should list the "removed files" in the "upgrade" section. [19:14]
MichaelDaumthere are new files as well [19:14]
gac410Hopefully any sane upgrade will get the new stuff. I worry more about stale stuff, sometimes they cause big issues, obsolete templates, etc. [19:15]
MichaelDaumfeels like window$ registry [19:16]
gac410:) [19:16]
MichaelDaumI just checked the state of JQueryPlugin on the other project [19:17]
gac410Well I'll get the test release built - 1.1.8_001 and will install it on foswiki.org [19:17]
MichaelDaumkool [19:17]
gac410As soon as our major languages are complete, I'll build RC1 ... hopefully never to see RC2
How is JQ on the "other project"
[19:18]
MichaelDaumtheir JQueryPlugin is far behind the curve in some state before we added it to the core
remember they don't have addToZone, nor %JQREQUIRE and all of this infrastructure. a lot of modules we ship with the core aren't part of TWiki. they have a separate plugin for the color formfield type. no textboxlist formfield. no rating formfield.
I checked the code quality of some of the newer plugins, i.e. PercentCompletePlugin.
Pth has no clue of jQuery.
none of their javascript code is able to run in no-conflict mode.
[19:20]
gac410Cool - Portuguese was just completed too. 6 languages to go.
Kicking off Foswiki Suite, for one last run, then the TestWeb sanity tests and I'll build / upload to f.o
I suppose I ought to perlbrew and run the suite on 5.8.8 and blead as well :P
[19:24]
MichaelDaum5.10.0 is the oldest Redhat and Suse Enterprise ship
unicode starts at 5.12.1
[19:28]
gac410We still advertise 5.8.4 I think, trunk was "featured" to move to 5.8.8.
I'll make a to-do note on my stone tablet
[19:29]
MichaelDaumthe only way out for those installs with such an old perl is perlbrew/plenv
if we happen to default to charset=utf8 ever, then 5.12.1 is a minimum requirement ... which isn't shipped anywhere as far as I've checked last time on distrowatch
[19:30]
gac410Even 5.10.0 is pretty bad. Shipping a "dot zero" as stable? 5.10.1 would have been better, and would also have gotten us by the version issue [19:31]
MichaelDaumlots of corps don't install the next service packages [19:32]
gac410yeah. I used to believe that, but as things move faster, there is *always* another bug, and staying on the edge I found was more often better than backlevel. [19:33]
MichaelDaumI had one case where their datacenter monitoring tool failed on newer perls as they updated. instead of fixing it they rolled back the virtual image backup.
this is when the only solution is to use perlbrew or plenv
just for foswiki
[19:33]
gac410Never tried plenv We ***really*** need to figure out our CPAN strategy for 1.2. The prereqs used by LogDispatch, which triggered Sven to pull into core too, are ~~difficult~~
As soon as a cpan prereq requires XS code, we are toast, can't distribute the .pm
[19:35]
MichaelDaumas is the case for ImagePlugin [19:35]
gac410That's the root cause of the nightly test fail on trunk. Perl prereqs [19:36]
MichaelDaumI've put them all in one script that runs cpanm on them [19:37]
gac410I'd really prefer for us to have a "cpanm" procedure to install to a local perl lib directory, [19:37]
MichaelDaum+1
configure then would be able to install missing reqs
[19:37]
gac410I started to work on a cpanm contrib that could do that, but got distracted. It's only going to get worse though. I'd love to nuke the lib/CPAN/lib directory [19:37]
MichaelDaumzombie land [19:38]
gac410Several folks argued that many sites don't allow that. The package manager or bust.
Really vehement against using cpanm or anything other than apt / rpm ... and when a distro doesn't package what we need, then what.
[19:38]
MichaelDaumwe could move the CPAN into a contrib of its own
so people could opt out
[19:39]
gac410Yeah, That's what I'd like, but a contrib that ships a list of packages, and a cpanm bootstrap. [19:39]
MichaelDaumthere's too much politics in distros and their package inventory [19:40]
gac410The objection there was all the servers that don't have internet connectivity. [19:40]
MichaelDaumhow would they get Foswiki-xxx.zip anyway
same channel is open for a CpanContrib
[19:40]
gac410Internal transfer via trusted paths, as opposed to direct internet connection. [19:40]
MichaelDaumoh we have a CpanContrib and CpanInstallerContrib already [19:41]
gac410There have been previous attempts, of varying degrees of success.
One used cpanplus, and I think that's been deprecated by upstream
cpanminus FTW in my book
[19:41]
MichaelDaumcpanm is the current kid on the block [19:42]
gac410yup
it "just works"
[19:42]
MichaelDaumawesome software
from what I see the only way to be real about unicode is to promote local perl more for foswiki
that way people have a way out of the rabbit hole without putting other perl systems in danger that are also running on the same server
[19:42]
gac410y, perlbrew is awesome too. another "it just works", like cpanm [19:45]
MichaelDaumone problem with perlbew and plenv is shebang [19:45]
gac410tools/rewrite-shebang works very well. I use it all the time [19:46]
MichaelDaumme too
but I just got bitten when releasing a new plugin ... and forgot to normalize the shebang again
[19:46]
gac410I think the FW package installer rewrites too during the install.
Been a while since I looked at it.
If it doesn't, it should :D
[19:46]
MichaelDaumit does just fine. no worries. [19:47]
gac410good! [19:47]
MichaelDaumit actually is the only sensible solution we have atm. [19:47]
gac410:) [19:47]
MichaelDaumthe other would have been #!/usr/bin/env perl
but that throws taintedness out of the fence
[19:48]
gac410Sven was looking at trying to make foswiki core installable via the extension installer, iirc. [19:48]
MichaelDaum... which might be a move we probably have to follow anyway for locales
jast found out that either you use -T _or_ use locale; both isn't going to work in perl.
[19:48]
gac410yeah, I recall.
somewhat stupid, there ought to be a way to untaint the locales
[19:50]
MichaelDaumhe said there is none
or incredibly complicated
[19:50]
gac410yeah. which is really stupid on behalf of perl core [19:51]
MichaelDaumI wonder how much we really loose when we disable taintedness by default [19:52]
gac410well, probably gain some perf, and I doubt it helps core much, as we hopefully find all the taint issues during dev.
Very few crop up on support
[19:52]
MichaelDaumrecent security issues were of different nature where taint checks played no role [19:53]
gac410right. once in a great while, someone will report a taint issue, and it's usually plain bug as opposed to a real issue
temp file path was a real pain.
[19:54]
MichaelDaumand then it is blindly untainted. mission completed. [19:54]
gac410:)
hopefully not
[19:54]
MichaelDaumback to square one: locales. they are only required to get sorting right and the like.
like the recent report about the wrong sorting in TablePlugin
[19:54]
gac410well lets get 1.1.9 out, and **finally** put a stake in the beast and get moving on 1.2. And locales.
and I18N in general
[19:55]
MichaelDaumlocales is 2.0 stuff my guts say [19:56]
gac410Though this time I am NOT going to say that 1.1.9 is the last 1.1. release. I've done that with 1.1.7 and 1.1.8, and been wrong twice :D [19:56]
MichaelDaum1.2.0 will be the first release I can be really happy with
1.2.0 will also require more IE9+10 fixes
[19:56]
gac410I can't help with IE, My only test systems are XP :P [19:58]
MichaelDaumgot a set of images running in virtualbox [19:58]
gac410cool [19:58]
MichaelDaumjust for the IEs [19:58]
gac410We need pharvey to resurrect selenium too.
IIRC Sven was somewhat interested in Bootstrap
[19:59]
MichaelDaumhe is interested in a lot of stuff
Bootstrap is pretty hype atm
however people had problems to adopt the latest major BS release as it is incompatible with the one before ... ever moving sands
[20:03]
gac410forward compat issues ... sounds a bit like a lot of the jquery extensions [20:05]
MichaelDaumI have to pack things up too. see you around. [20:06]
gac410Are the removed jquery files going to cause any issues for anyone with local apps? [20:06]
MichaelDaumnot that I could imagine [20:06]
gac410okay great [20:07]
................................. (idle for 2h44mn)
pharveyI shall endeavour to resurrect selenium stuff. Not sure how soon.. I've a big work task that is dragging on and on [22:51]
gac410no problem. I've never gotten it working. We'll probably appreciate it once we need to work on 1.2 / 2.0 and IE 9 / 10 support [22:59]
Damn... TMCE / Wysiwyg plugin is inserting &#194 into a test document for some reason.
I'm seeing <p class="WYSIWYG_NBNL">&#194;&#160;</p> and a simple <p>&#194;&#160;</p> showing up in the tmce html view
wtf. I've never seen that before.
[23:11]
....... (idle for 32mn)
Something definitely broken on 1.1.9 / Release01x01 checkout.
Create a new empty topic, put cursor on start of an empty line. Click format -> Heading 1. Line fills with lots of special characters.
TML topic - 3-4 blank lines between TML. Click wysiwyg, view html, blank lines are encoded as &#194;&#160; On save, the &#160; is removed, but 94 remains as a strange character.
[23:46]

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