Release Meeting: 20 April 2015


1. Status update on Translations

  • Translations for Czech, Danish, French, German and Italian are in great shape.
  • We need help enlisting translators for other languages. ... Ideas?
There was some discussion of translations. We'll try another email, but we may just have to drop or ship the translations as incomplete. There was also some discussion of possibly releasing the translations as a separate package. Because the translations are tightly tied to the strings within the topics & code, we'd need to tightly bind the translations with the foswiki release. We really don't have any infrastructure available to do that.

2. Release blockers

12 "Blockers",

Item13331 needs some special consideration

Rendering of Czech form rendering is broken due to entity encoding issues in CGI. Item13331 branch starts to convert generation of HTML from using CGI::... methods, over to using templates. However it's turning into a redesign and work should really be a feature request. Alternatives:
  • Delay 1.2, and go back and design this correctly.
  • Release 1.2 as-is with broken rendering of Czech.
  • ???

Excellent progress was made by MichaelDaum on this task. There is a CGI configuration setting that gets around the issue. The work on the Item13331 branch will be tabled. There are numerous reasons to not try to rush the conversion to templated based HTML into 1.2. It really needs a feature proposal and careful design. It buys us nothing to hard-code the html into the code. We need to move class assignments, into the templates, for ex.

Item13331 was closed shortly after the meeting

Other blockers

  • Configure tasks:
    • 13364 (Closed): Very bad user experience with the new configure - installing extensions - probably ready to be closed
    • 13359 (No Action Required): Crash in Configure/ after installing extension - probably can be closed - fixed under other tasks.

  • New and escalated
    • 13369 (Closed): Wysiwyg converts entities to unicode in HTML2TML, and corrupts MAKETEXT strings.
    • 13371 (Closed): Encoding problem on utf8 VERBATIM block
    • 13372 (Closed): Encoding problems (utf8) on Macro arguments

These issues were closed shortly after the meeting

  • Being worked on
    • 8966 (Confirmed): Wysiwyg editor breaks standard topics Topics being reviewed for issues - Need help, it's time consuming to edit each topic and then compare the results.

  • Not really blockers
    • 12855 (Closed): Audit core DEPENDENCIES - Admistrivia / Documentation update for when we are ready to build release packages.
    • 11262 (Closed): Update Installation Guide - We can solicit review from users testing the beta packages.
    • 13139 (Closed): Backwards compatibility issues with Foswiki 1.2 default extensions - Part of dependency review / testing effort. Need to test all 1.2 extensions against 1.1 (and 1.0.x?)
    • 12888 (Closed): Unit tests are failing - Catch-all. Currently no tests failing.
    • 12246 (Closed): Documentation for PageCaching is out of date - Cache really needs a documentation review

A new blocker was opened late on Monday. 13378 (Closed): Incorrect rendering utf8 WikiWords in the Form's default text., certain utf8 characters are being linked as wiki words when used in a form.

3. Proposed Schedule for Alpha / Beta / GA

  • Beta 1: Built on Friday April 3rd. delayed due to blockers.
  • Beta 2: (Need to decide on when to build next beta)
  • Release Candidate 4/18 (Translation complete) If no checkins in 24 hours, RC becomes release?

Next beta will be later this week

Next actions

Next meeting - - Monday May 5, 2015 1300Z


   (08:46:46 AM) Lynnwood  entered the room.
   (09:01:37 AM) gac410: Hi everone. Agenda:
   (09:01:50 AM) CDot  entered the room.
   (09:02:36 AM) JulianLevens  entered the room.
   (09:03:01 AM) MichaelDaum: welcome on board
   (09:03:26 AM) gac410: yes good day everyone Or is that good night for mtempest ?
   (09:03:53 AM) CDot: gac410: mtempest is in RSA (unless heʼs moved)
   (09:04:00 AM) mtempest: 3om
   (09:04:05 AM) CDot: so roughly a European TZ
   (09:04:17 AM) CDot: mtempest: so where are you now?
   (09:04:17 AM) mtempest: Er, 3pm here in South Africa
   (09:04:21 AM) gac410: Ah... okay so good day everyone :)
   (09:04:32 AM) CDot: ah, I thought so - 3am had me guessing ;-)
   (09:04:58 AM) CDot: gʼday George
   (09:05:19 AM) ***mtempest waves
   (09:05:29 AM) gac410: I was hoping we were getting close to Beta 2 but the recent testing mainly by jomo has opened up a can of worms. :(
   (09:05:55 AM) jast: first things first... translations?
   (09:05:56 AM) gac410: But first on the agenda ... Translators. Any suggestions on getting help for the languages that are being left behind?
   (09:06:53 AM) CDot: gac410: which languages are most pressing?
   (09:07:14 AM) gac410: Spanish is one that was 100%
   (09:07:31 AM) gac410: r Czech, Danish, French, German and Italian are in great shape.
   (09:08:08 AM) ***CDot knows several Spanish speakers, but none who would be able to translate :-(
   (09:08:46 AM) gac410: If you look at release 1.1 had a number of languages at 100%, that are in the 40-60% on master
   (09:10:09 AM) gac410: Korean, Norwegian, Polish, Portuguese (brazil) Russian, Spanish, Swedish, Traditional Chinese .... were all in the high- 80% to 100% range on 1.1, and are not making much headway on 1.2
   (09:12:47 AM) gac410: We can do another plea to foswiki-announce & foswiki-discuss ... I've tried once or twice. Anyone else want to take a stab at it?
   (09:13:29 AM) ***CDot thinks there probably isnʼt much point
   (09:14:08 AM) CDot: we depend on the languages spoken by the people who are engaged at the time
   (09:14:25 AM) CDot: trying to engage someone new just for translations..... I doubt it
   (09:15:17 AM) gac410: So we can drop the languages from the MANIFEST or just release them in incomplete status. Someone suggested releasing a LanguagesContrib Was that you CDot?
   (09:18:07 AM) CDot: y
   (09:18:37 AM) CDot: the issue with that is not being able to strongly like versions of contribs with releases of FW
   (09:18:43 AM) CDot: ^like^link
   (09:19:13 AM) CDot: as in "you need version 6 of GermanContrib for version 1.3.0 of Foswiki"
   (09:19:32 AM) CDot: "but version 2 for 1.1.9" (etc)
   (09:20:06 AM) gac410: Right. I think this is one to sidestep for now :) Especially if we try to release more often with a continuous development stream
   (09:20:33 AM) CDot: for now, yes, but with continuous integration we need something like that.
   (09:20:51 AM) mtempest: Have a separate contrib for each foswiki version?
   (09:21:07 AM) CDot: mtempest: yuck
   (09:21:41 AM) CDot: perhaps a git branch for each FW version.....
   (09:21:42 AM) gac410: We just are not structured to have tight coupling in Extensions. There have been proposals in the past to fix it, but it is *major* surgery I suspect.
   (09:22:21 AM) JulianLevens: I have started looking into this sort of thing and CI but itʼs really early days
   (09:22:40 AM) gac410: If we "always release from master" and branch for features, and merge into master for release, I don't think we need to worry as much about the languages.
   (09:22:59 AM) gac410: Keep the tip of Master in a "ready to release" state.
   (09:24:10 AM) JulianLevens: That suggest that a CI would look for Tasks ʼWaiting for Releaseʼ and automatically merge into a daily_build branch well er daily
   (09:24:29 AM) JulianLevens: Run the tests and report problems
   (09:24:47 AM) JulianLevens: If all tests pass then merge them to master?
   (09:25:24 AM) ***gac410 doesnʼt know if we are big enough to try to tackle full automation of merging.
   (09:25:40 AM) gac410: I think it would be better to make manual decisions on what merges into master.
   (09:26:03 AM) JulianLevens: Neither am I, but Iʼd like to see how far I can get
   (09:26:11 AM) gac410: :)
   (09:26:56 AM) gac410: I think we need a separate meeting / proposal on how we will move forward with git branch/merge strategy. ... lets get 1.2 out first :)
   (09:27:22 AM) gac410: Well let's move on from translations.
   (09:28:01 AM) gac410: Release blockers. First the big one Item13331 ... big issues in Forms with certain codepoints.
   (09:28:39 AM) jast: is this still an issue with CGI 4.15?
   (09:28:43 AM) jast: (the bug report mentions 4.14)
   (09:28:53 AM) jast: there have been changes to how entities are processed
   (09:29:01 AM) gac410: It seems to be a CGI bug. Old versions of CGI work fine if the Language is set. Hm. I didnt' know about 4.15 must have just come out.
   (09:29:11 AM) jast: released today
   (09:30:03 AM) jast: it seems like it now only encodes these: <>&"&#700;
   (09:30:08 AM) jast: and two extra special ones
   (09:30:26 AM) jast: plus you can now configure the entity processing using a global variable in CGI
   (09:31:13 AM) CDot: gotta hate CGI :-(
   (09:31:31 AM) CDot: it has always been a thorn in our sides.
   (09:32:16 AM) MichaelDaum: any html rendering is going to be removed from CGI-5
   (09:32:34 AM) MichaelDaum: so Foswiki::Render::HTML is the RTTD
   (09:32:35 AM) gac410: If someone can test CGI 4.15 on master, There are testcases attached to
   (09:33:16 AM) MichaelDaum: interestingly 4.14 is missing from
   (09:34:21 AM) MichaelDaum: CGI::Textareas are still broken on for us using a DataForm field textarea containing Nech&#357; již h&#345;íšné saxofony &#271;ábl&#367; rozezvu&#269;í sí&#328; úd&#283;snými tóny waltzu, tanga a quickstepu.
   (09:34:39 AM) gac410: On Item13331 or on master?
   (09:34:59 AM) MichaelDaum: master
   (09:35:12 AM) gac410: CGI version?
   (09:35:13 AM) MichaelDaum: google translate says: Now let sinful saxophones devils chime hall appalling tones waltz, tango and quickstep.
   (09:35:16 AM) MichaelDaum: 4.15
   (09:35:23 AM) MichaelDaum: just cpamn&#700;ed it out of the net
   (09:36:19 AM) MichaelDaum: it shows "Nech&#357; již h&#345;íšné saxofony &#271;ábl&#367; rozezvu&#269;í sí&#328; úd&#65533;›snými tóny waltzu, tanga a quickstepu." in edit
   (09:36:38 AM) MichaelDaum: &#3848; &#3921;&#3904;&#3938;&#3851;&#3928;&#3931;&#3962;&#3942;&#3851;&#3944;&#3851;&#3937;&#3954;&#3906;&#3851;&#3939;&#3942;&#3851;&#3936;&#3905;&#4018;&#3956;&#3908;&#3942;&#3851;&#3937;&#3962;&#3851;&#3940;&#3962;&#3942;&#3851;&#3906;&#3919;&#3962;&#3938;&#3853; &#3853;&#3925;&#3942;&#3851;&#3938;&#3986;&#3964;&#3939;&#3851;&#3933;&#3851;&#3942;&#3984;&#4017;&#3962;&#3942;&#3851;&#3935;&#3954;&#3939;&#3851;&#3906;&#3923;&#3964;&#3923;&#3851;&#3906;&#3921;&#3964;&#3908;&#3851;&#3939;&#3988;&#3851;&#3926;&#3934;&#3954;&#3923;&#3853; &#3853;&#3910;&#3906;&#3942;&#3851;&#3920;&#3964;&#3906;&#3942;&#3851;&#3904;&#3956;&#3923;&#3851;&#3926;&#4018;&#3939;&#3851;&#3928;&#3930;&#3956;&#3908;&#3942;&#3851;&#3928;&#3962;&#3921;&#3851;&#3936;&#3911;&#3928;&#3851;&#3926;&#4017;&#3908;&#3942;&#3851;&#3928;&#3920;&#3956;&#3942;&#3853; &#3853;&#3928;&#3851;&#3943;&#3953;&#3851;&#3928;&#3905;&#3942;&#3851;&#3924;&#3936;&#3954;&#3851;&#3906;&#3929;&#3964;&#3851;&#3926;&#3964;&#3851;
   (09:36:40 AM) MichaelDaum: &#3913;&#3954;&#3921;&#3851;&#3906;&#4017;&#3956;&#3938;&#3851;&#3909;&#3954;&#3906;&#3853; is even worse
   (09:36:40 AM) gac410: yup That's what happens with Czech e-Caron
   (09:38:23 AM) MichaelDaum: maybe that&#700;s these two extra codepoints that CGI still tries to translate calling HTML::Entities
   (09:39:03 AM) MichaelDaum: we could try to provide a custom CGI::ENCODE_ENTITIES
   (09:39:16 AM) MichaelDaum: defaults to &<>"\x8b\x9b&#700;
   (09:40:50 AM) gac410: Michael, if you and jast could work on master, and try to get this all working with CGI, I'll create a feature proposal to cover Item13331 and will try to work on that branch in parallel.
   (09:41:24 AM) gac410: We need to decide what to do with the rest of the blockers, mainly Wysiwyg now :(
   (09:41:50 AM) gac410: And can make a later decision on whether to accept 13331 into master, or defer it to a post 1.2 release.
   (09:42:24 AM) gac410: I think configure is now in great shape. CDot has done a masterful job :D
   (09:42:50 AM) CDot: I have been looking at the MAKETEXT report you came up with, and at this point in time I&#700;m not sure it&#700;s fixable
   (09:43:01 AM) CDot: at least not without hacking TMCE
   (09:43:48 AM) gac410: There is one lurking bug, in javascript. with extension installer, *sometimes* the args={} hash is empty after selecting one or more extensions to install. But works on second attempt.
   (09:44:11 AM) gac410: CDot: it fails in the ROUNDTRIP Wysiwyg test, is TinyMCE even in the loop?
   (09:44:12 AM) CDot: don&#700;t give me "sometimes". I don&#700;t do "sometimes" :-(
   (09:44:42 AM) CDot: gac410: the ROUNDTRIP test is flawed; I can get that test to pass, but when you put TMCE in the loop, fur flies.
   (09:45:21 AM) CDot: I will keep working on it, but can&#700;t promise a fix.
   (09:46:09 AM) gac410: ah foo. That sucks. I guess best bet for 1.2 is to <sticky> them then. Which I already did.
   (09:46:09 AM) gac410: re "sometimes" ... yeah I know cdot, If you use the Search, select one or two extensions, and click report, or install, on occasion you get just a dialog with the word "Validate" ... nothing else.
   (09:46:36 AM) CDot: when you have details of how to reproduce it, shout.
   (09:46:38 AM) gac410: I traced the POST, and when that happens, the args={} hash is empty, as if nothing was checked. Repeat, and it works.
   (09:46:43 AM) MichaelDaum: $CGI::ENCODE_ENTITIES = q{&<>"&#700;}; fixes it
   (09:47:46 AM) gac410: Ah. great. So lets make that the solution for 1.2! (And you may need to force CGI Charset to utf-8, I found it using iso-8859-1 even when utf-8 configured for foswiki.
   (09:47:57 AM) gac410: That fixes older CGIs setting the right charset
   (09:48:18 AM) gac410: Those details are in the 13331 task
   (09:49:28 AM) ***MichaelDaum not sure where to put $CGI::ENCODE_ENTITIES = q{&<>"&#700;}; strategically
   (09:49:46 AM) MichaelDaum: I&#700;ve added the patch for Foswiki::Form::Textarea to
   (09:49:59 AM) MichaelDaum: I&#700;ll forward our findings to the CGI maintainer
   (09:50:29 AM) jast: in CGI 4.15 the CGI charset is no longer used for HTML escaping
   (09:50:54 AM) gac410: Right, but most sites are running older versions. 3.65, for eg.
   (09:50:59 AM) jast: and even in 3.x it affects only special cases
   (09:51:50 AM) jast: in 3.x we should be safe if we set it to empty string
   (09:52:10 AM) jast: the only values that make a difference for HTML escaping are: ISO-8859-1 and WINDOWS-1252
   (09:52:25 AM) jast: (yes, that makes little sense, but it&#700;s what the code does)
   (09:52:27 AM) gac410: The current state is that I'm getting lost. We have a lot of TranslationTests failing now. I don't think the patches attached to Item13372 are safe for general use.
   (09:52:49 AM) MichaelDaum:
   (09:53:44 AM) jast: anyone else get the impression that the maintainer of changed recently? :}
   (09:54:47 AM) MichaelDaum: no need for impressions. yes Lee took over after CGI being unmaintained for some time.
   (09:55:03 AM) MichaelDaum: he held a talk about CGI&#700;s future direction
   (09:55:19 AM) jast: he&#700;s probably learning a thing or two about the joys of being a maintainer of CGI :}
   (09:55:26 AM) gac410: The things we are struggling with are mostly stuff that is deprecated and will be eliminated.
   (09:55:47 AM) MichaelDaum: Lee is a sharp & nice guy
   (09:56:02 AM) gac410: Okay. Back to blockers and then Beta 2
   (09:56:27 AM) gac410: 13331, 13369, 13371 and 13372 are all Wysiwyg and/or CGI entity related.
   (09:56:30 AM) jast: so, we have an interim solution for 13331?
   (09:57:00 AM) MichaelDaum: jast, see the patch there. it needs to move though to a better place, maybe.
   (09:57:05 AM) gac410: 13331, yes. Set charset in older CGI, and override ENCODE_ENTITIES in newer
   (09:58:19 AM) MichaelDaum: gac410, your work isn&#700;t lost. please keep the branch.
   (09:58:43 AM) MichaelDaum: we will have to follow up on Foswiki-2
   (09:59:07 AM) gac410: Yes I will. I might rename it .. but definitely not discarding it. I'll open a feature request to match for a future release.
   (09:59:17 AM) MichaelDaum: xcellent
   (09:59:45 AM) MichaelDaum: given low-fi approach cures 13331
   (10:00:05 AM) gac410: Yes. We need to get this all into Beta 2, and get jomo and others to beat on it.
   (10:01:44 AM) gac410: So hurdles for Beta 2. Get the TranslatorTests passing and the 13331 fix in. All inter-related with encoding issues.
   (10:03:48 AM) gac410: CDot ... The 13369, 13371 and 13372? For at least MAKETEXT I think it's fine to declare that we use <sticky> to protect. But Item13372 is worrying.
   (10:05:16 AM) gac410: And 13371 as well :(
   (10:05:27 AM) CDot: They are all aspects of the same problem.
   (10:05:37 AM) gac410: Okay ... well that's good I guess :)
   (10:05:49 AM) CDot: no, because it&#700;s a very hard problem to solve
   (10:06:02 AM) MichaelDaum: said Turing
   (10:06:29 AM) CDot: "NP", said Turing, "It&#700;ll only take me 10^28 years to solve that"
   (10:06:42 AM) MichaelDaum: NM
   (10:07:32 AM) jast: some day someone&#700;s going to prove that P = NP
   (10:07:33 AM) CDot: anyway, if anyone wants an explanation of the problem, see me after class :-)
   (10:07:44 AM) MichaelDaum: "divide an conquer" is now "ignore and forget"
   (10:07:46 AM) jast: ... introducing a constant factor of 10^(10^100)
   (10:08:59 AM) MichaelDaum: if fixing WysiwygPlugin is NP-complete then "ignore and forget" is the right solution for Foswiki-1.2.0
   (10:09:42 AM) MichaelDaum: because infinity is very long, especially the last part
   (10:09:43 AM) CDot: no, it&#700;s not that sort of problem. The problem is that neither WysiwygPlugin, nor TMCE, are lossless
   (10:09:59 AM) jast: I think I know about 70% of the problem, and that&#700;s enough for me to not know how to solve it
   (10:10:15 AM) CDot: both lose information, and there&#700;s a delicate tradeoff about how much of a mess you make of the HTML that TMCE edits (to protect it from TMCE)
   (10:10:24 AM) CDot: and the editabilty of the HTML
   (10:11:07 AM) gac410: If we can at least *detect* that the HTML will not be reliably editable, we could just disable wysiwyg ... we do that for some other stuff I think.
   (10:11:16 AM) gac410: That would be okay for 1.2 IMHO.
   (10:11:27 AM) CDot: IIRC there is already a switch to do that
   (10:11:38 AM) gac410: Right, but can we automate this case?
   (10:11:39 AM) jast: if we can do that, I agree it&#700;s enough
   (10:11:49 AM) CDot: it turns off WYSIWYG if the test contains macros
   (10:11:58 AM) CDot: text
   (10:12:12 AM) gac410: Detect that the topic contains text that will not edit cleanly?
   (10:12:24 AM) CDot: yes, that what it does
   (10:13:29 AM) jast: if it contains *any* macros?
   (10:13:50 AM) CDot: can&#700;t remember the details. It&#700;s doccoed in the WysiwygPlugin topic.
   (10:14:06 AM) jast: right
   (10:15:39 AM) gac410: Anyway, CDot, your best judgement, if we can somehow prevent Wysiwyg from corrupting utf8, even if it means disable Wysiwyg if text contains certain entities, that's fine for 1.2
   (10:16:52 AM) gac410: 1.2 was not intended to be our "unicode release" The feature crept in once taint was disabled.
   (10:17:06 AM) CDot: it&#700;s not "corrupting UTF8". That particular problem was solved several days ago. What it&#700;s doing is being overenthusiastic to convert entities to codepoints
   (10:17:13 AM) CDot: and that&#700;s breaking macros.
   (10:17:27 AM) jast: so, add a check that disables WYSIWYG if problematic entities are in the text
   (10:17:35 AM) CDot: it works with iso-8859-1 because it *has* no codepoints outside of ascii (well, not many)
   (10:17:55 AM) gac410: okay.
   (10:18:00 AM) jast: IMO topics shouldn&#700;t be using entities in most cases anyway
   (10:18:19 AM) CDot: yes, you could add such a check, as a band-aid. As I said, I can&#700;t guarantee a fix.
   (10:18:43 AM) gac410: I was a little concerned that Wysiwyg is converting &quot to the numeric equivalent ... that will be confusing to users. But that's a different issue.
   (10:18:48 AM) jast: well, as long as we have something that can get the release out, the blocker is manageable, that&#700;s the main thing now
   (10:19:04 AM) gac410: It doesn't round trip any of the named entities any more ... I don't think.
   (10:19:28 AM) CDot: if you are using UTF8, there are codepoints for all the named entities
   (10:20:36 AM) CDot: anyway, I have a small testcase, but as I said, I can&#700;t guarantee a fix. Not because it isn&#700;t fixable - it is, I&#700;m sure - but because it&#700;s complicated and risky.
   (10:20:42 AM) gac410: Right. So a user editing a topic with &quot saves and then next time they edit (in plain text) they see &#xx; wtf? But again, it sounds like this is something we have to live with with utf8
   (10:20:50 AM) CDot: thank goodness we have a large test set for WYSIWYg.
   (10:21:11 AM) gac410: Where it impacted us was mainly ME ... I was wysiwyging release topics, and converted all the entities to codepoints :(
   (10:21:48 AM) CDot: gac410: please leave that one; it&#700;s really icing on the cake
   (10:22:06 AM) gac410: leave it?
   (10:22:11 AM) CDot: it&#700;s much more important to solve the conversion of entities (named or otherwise) to codepoints
   (10:22:27 AM) gac410: Right. I won't make that a blocker ... if that's what you meant.
   (10:22:28 AM) CDot: which is a completely different problem
   (10:23:03 AM) CDot: BTW, whether it uses named or numbered entities is up to HTML::Entities, so if you want, you can find a way to "explain" your concerns to it.
   (10:23:03 AM) gac410: I should probably open it as a low or normal, just to get it documented, but definitely not for 1.2
   (10:24:14 AM) gac410: So to make sure I understand. If using a site character set that has codepoints for named entities, Wysiwyg will convert the named entity to the equivalent codepoint.
   (10:25:23 AM) gac410: Okay .. .back to release blockers. Anything else? Any new tasks we should escalate?
   (10:26:25 AM) gac410: Lynnwood has a new draft InstallationGuide. if anyone else wants to review?
   (10:26:36 AM) CDot: IIRC TMCE converts *everything* to numbered entities, which is where the problem starts.
   (10:26:36 AM) mtempest: Is that not configurable?
   (10:26:36 AM) CDot: mtempest: possibly; I&#700;ve been avoiding the TMCE JS like the plague
   (10:26:36 AM) mtempest: :-)
   (10:26:36 AM) ***CDot has his hands full with WysiwygPlugin
   (10:26:36 AM) CDot: there&#700;s a new version of TMCE queued, which changes all sorts of APIs, which is one good reason not to invest time on it.
   (10:26:36 AM) gac410 left the room (quit: Ping timeout: 240 seconds).
   (10:26:36 AM) The account has disconnected and you are no longer in this chat. You will automatically rejoin the chat when the account reconnects.
   (10:27:48 AM) gac410: Oops. my connection dropped.
   (10:28:14 AM) gac410: Last thing I saw: CDot: there's a new version of TMCE queued, which changes all sorts of APIs, ...
   (10:28:28 AM) CDot: gac410: you didn&#700;t miss anything
   (10:28:33 AM) gac410: okay thanks.
   (10:30:04 AM) gac410: Okay, so Beta 2 ... I'll plan to build it as soon as CDot is comfortable with the Wysiwyg fixes. And jast / jomo / MichaelDaum verify the CGI 4.15 fixes.
   (10:30:55 AM) MichaelDaum: verified and checked in. need somebody to x-check
   (10:31:14 AM) gac410: I'll open tasks on entity -> codepoint conversion as a low priority, and Configure extension installer wizard, as intermittent ... needs documentation and a js expert to figure out why configure.js is occasionally not setting the args hash.
   (10:31:29 AM) gac410: Neither of which will block Beta 2.
   (10:33:29 AM) gac410: Do we know of anyone other than jomo that is really testing the Beta ?
   (10:33:56 AM) gac410: (outside of the core dev's that is :) )
   (10:34:11 AM) Lynnwood: i&#700;ve been doing some testing on it.
   (10:34:34 AM) Lynnwood: got a bit distracted in trying to install plugin which i found out was broken.
   (10:34:38 AM) gac410: Ah yes sorry Lynnwood ... I should have remembered that.
   (10:34:45 AM) Lynnwood: i&#700;ll go back and go through some more basic testing.
   (10:34:56 AM) Lynnwood: no worries
   (10:35:56 AM) Lynnwood: after i found plugin install was broken, i went to see about getting a newer copy from git and discovered that was more involved than first thought.
   (10:36:20 AM) gac410: And one more issue. in yesterday's irc log, user "foswiki_irc8" could not get attach to work. Kept getting unable to write the CGI temp file.
   (10:36:24 AM) Lynnwood: so i&#700;m going to go back and do fresh install of the beta and resume basic testing.
   (10:37:07 AM) gac410: Looking back through the logs, only one other user - 4 years ago - had the same issue. I had him patch in the 1.2.0 fixes to override temp file locations, but still he was unable to get upload to work.
   (10:37:14 AM) gac410: Lynnwood: okay great. Thanks.
   (10:38:52 AM) gac410: Even went as far as to set the various TMP env. variables in setlib.cfg and it still failed with permission denied. I have no idea and user was going to ask his hosting provider, and then planned to move to a different wiki software :(
   (10:40:05 AM) gac410: Anyway, I think we are in pretty good shape for Beta 2.
   (10:40:34 AM) MichaelDaum: in the meantime Lee reopened
   (10:40:39 AM) CDot: oh, crap. i just realised I have already fixed 13369
   (10:40:50 AM) gac410: Any further business. If not Thanks everybody for all the hard work!
   (10:41:02 AM) MichaelDaum: thanks to you gac410.
   (10:41:11 AM) CDot: yes, thanks George
   (10:41:43 AM) Lynnwood: big thanks to you gac410 !
   (10:41:45 AM) ***gac410 is good at shoveling work on CDot and MichaelDaum and others. :D
   (10:41:53 AM) jast: yes, major thanks for filling the unenviable role of making hard calls and running after everyone :)
   (10:42:41 AM) gac410: We'll get this code out eventually. Then we can reset, and figure out our branching / build / release strategy going forward ... and start to prioritize work for the next great release !
   (10:43:24 AM) gac410: After playing with the item branch ... I really think a branch/merge strategy is going to make this all much easier to keep a clean master "ready to release"
   (10:43:33 AM) jast: absolutely
   (10:43:37 AM) JulianLevens: y
   (10:44:27 AM) gac410: Oh... one question. We are releasing FCGI and ModPerl with 1.2. but I never restructured the repositories into distro. I think I still should do that. ?
   (10:48:41 AM) MichaelDaum: y
   (10:49:59 AM) CDot: BTW i keep getting "mod perl not installed" errors from the installer. is that expected>
   (10:50:42 AM) gac410: yes. Since we bundle the extension, it should be an "optional" dependency, only needed if you intend to use it.
   (10:50:46 AM) CDot: maybe they are warnings. perl developer
   (10:51:11 AM) gac410: Ah yeah pseudo-install doesn't do a very good job of dependency management. Everything is an error.
   (10:51:17 AM) CDot: it&#700;s a bit dosconcerting, it has to be said.
   (10:51:30 AM) CDot: disco-certing
   (10:51:46 AM) gac410: Well when I run it I get an error for pretty much every dependency
   (10:52:33 AM) gac410: MichaelDaum: leejo seems to imply with his last message that it's a windows issue???
   (10:54:21 AM) MichaelDaum: he doesnt seem to be 100% sure. promised to find some ol&#700; windows vm for testing as well as remove those hex codes from ENCODE_SHMATITIES in the next release of CGI
   (10:55:53 AM) gac410: MichaelDaum: regarding the ENCODE_ENTITIES ... I think that maybe it, plus setting of the CGI::CHARSET for older CGI should be done in, maybe around line 2044, and move $Foswiki::cfg{Site}{CharSet} ||= 'iso-8859-1'; to the same place.
   (10:56:03 AM) gac410: Set them just before creating the new response object.
   (10:56:38 AM) MichaelDaum: whatever fixes the beast
   (10:57:06 AM) MichaelDaum: I&#700;ve put ENCODE_ENTITIES into as it seems to be the very first code to "use CGI"
   (10:58:14 AM) gac410: Trying to figure out where to place stuff is painful. Ah... I don't know why, but I thought it was in Configure::Load. I see that now.
   (11:01:48 AM) mtempest: Cdot, I&#700;m now working at Airbus Defence and Space Optronics South Africa.
   (11:02:46 AM) CDot: mtempest: if it wasn&#700;t for the "defence" bit of that, I&#700;d say "sounds a lot safer" ;-)
   (11:03:11 AM) CDot: "Space Optronics" sounds like fun
   (11:03:21 AM) gac410: MichaelDaum: I'll move the CharSet handling around just a bit in and add a call to CGI::charset($Foswiki::cfg{Site}{CharSet}); which should fix older versions of CGI
   (11:03:47 AM) MichaelDaum: yay
   (11:04:08 AM) mtempest: Cdot it has its moments :-)
   (11:04:44 AM) mtempest: But I don&#700;t do space stuff
   (11:05:33 AM) gac410: And will also add $CGI::TMPDIRECTORY = $Foswiki::cfg{TempfileDir}; to the temp file setup... not that it seemed to fix foswiki_irc8's upload issues.
   (11:16:17 AM) gac410: mtempest: Sounds like interesting work.
   (11:17:12 AM) gac410: Okay MichaelDaum jast, ... I've tested jomo's form issues with cgi 3.65, and it seems to be clean.
   (11:17:45 AM) mtempest: gac410 :-)
   (11:20:22 AM) MichaelDaum: yes it is fine now
   (11:21:12 AM) gac410: phew... So I need to update the checker. Do we need to flag any CGI 4.11 -> 4.14?
   (11:31:00 AM) MichaelDaum: 4.14 is okay. frankly I don&#700;t know what the diff between 4.14 and 4.15 is. the change logs even dont mention 4.14. so I assume it is a minor glitch.
   (11:38:17 AM) mtempest left the room (quit: Quit: Bye).
   (11:38:37 AM) jast: 4.11 through 4.13 are bad and unfixable
   (11:38:54 AM) gac410: MichaelDaum: reviewing the changes it seems cosmetic 4.14 to 4.15
   (11:39:07 AM) JulianLevens left the room.
   (11:39:42 AM) MichaelDaum: okay
   (11:39:44 AM) MichaelDaum: jast, yes.
   (11:40:55 AM) jast: 4.11 was released in December, so there&#700;s a decent chance people are actually affected by one of these versions

Topic revision: r3 - 21 Apr 2015, GeorgeClark - This page was cached on 06 Aug 2020 - 10:59.

The copyright of the content on this website is held by the contributing authors, except where stated elsewhere. See Copyright Statement. Creative Commons License    Legal Imprint    Privacy Policy