#foswiki 2012-04-02,Mon

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

WhoWhatWhen
SvenDowideitgac410, ignore win8 unless someone has it on a released platform
i'd love things to be fixed before release, but really, win8 is er, a long time away
strikeone is what i should work on, but i don't see me figuring it out
but as my use is a 'new and innovative one' >sarcastic snigger<
the strikeone issue happens (now) on every foswiki i try it on
[00:11]
gac410well unless someone is going to pull a miracle in the next 12 hours or so, we'll have to delay 1.1.5 :( [00:14]
SvenDowideitjust for the tables? [00:15]
gac410I don't know. I'm tyring to figure it out but the wysywyg stuff is so complex. :(
I think I agreed with pharvey that we could do a 1.1.5 release without the table fix. But i'm pretty unsure.
[00:15]
SvenDowideitmmm, so thats a corner case where the experienced foswikiers use %BR% [00:17]
gac410It definitely doesn't fail on 1.1.4 - just tested. It's a new 1.1.5 bug either in TMCE or the WYSIWYG plugin [00:18]
SvenDowideityay :)
pharvey, can we steal an hour or 2 of your morning?
Item11312 is a sucky snotnose
[00:19]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [00:19]
pharveyd'oh. [00:20]
gac410For some reason the blank line causes (something?) to terminate the previous table. Not sure if it's down in the TML2HTML parser, or ... somewher eles [00:20]
SvenDowideitonly because, as gac410 points out
it was ok on 1.1.4
step 1
do we have a unit test?
[00:20]
pharveyI began by fixing selenium, but that's still broken. [00:20]
SvenDowideiti can start looking at that rabbit hole, but i've brain issues [00:20]
gac410No. I can't tell if it's in the WYSIWYG or TMCE side of things. [00:20]
pharveyI guess we could ignore selenium for now. [00:21]
SvenDowideityup [00:21]
pharveyIt's WYSIWYG. [00:21]
SvenDowideitstart with a in perl unit test [00:21]
gac410Oh... good. So that can be tested independent of the JS [00:21]
pharvey"It's WYSIWYG" -> my instinct so far, which could be wrong :) [00:21]
SvenDowideitok, drop it [00:21]
pharveyI'll add the test now to confirm. [00:21]
SvenDowideiti'll write the test [00:22]
gac410Something is "closing" the table... the blank line causes WYSIWYG to emit the /td /tr /table from what I can tell. [00:22]
SvenDowideitpharvey, i presume you have real work to do? [00:22]
pharveyyeah :( [00:22]
SvenDowideiti'll do it
and call you if i get stuffed
[00:22]
pharveykool [00:22]
gac410I'll help however I can [00:22]
SvenDowideitgac410, you always do :p [00:22]
pharveySvenDowideit, gac410 - I am curious if this was introduced in Foswikirev:9886 or Foswikirev:10527 [00:24]
FoswikiBothttp://trac.foswiki.org/changeset/9886 [ Changeset 9886 – Foswiki ] http://trac.foswiki.org/changeset/10527 [ Changeset 10527 – Foswiki ] [00:24]
SvenDowideitpharvey, what unit test file does it go into?
i do't speak the english that cdot does, so io-dunno
[00:24]
pharveyI'd put it in TranslatorTests.pm, but I'm not sure it matters [00:25]
SvenDowideitroger
its more a q of 'what do i copy from' :)
[00:25]
gac410I wish WYSIWYG had the trace code similar to the SafeWikiPlugin ... I was starthing to look down into the leaf and node stuff. That's where the SafeWikiPlugin stuffs up the html. the premature close of the table sounds suspiciously like some of the SWP issues. [00:26]
SvenDowideitpharvey, is planning on re-writing the entire parser thing next week :/ [00:27]
pharveyhaha. Actually, I'm really looking forward to that :P [00:27]
SvenDowideiti'm looking forward to zero code [00:27]
pharveyleaning on Marpa seems like such a cool idea. [00:27]
SvenDowideitthat would be a dream
but also means compiler platofmrs
making cpan/deb/rpm even more important
[00:28]
pharveyit has a pureperl version [00:28]
SvenDowideitwhich is much slowerer? [00:29]
gac410MARPA: Modification and Replacement Parts Association (I guess that's not it :D ) [00:29]
pharveywell, we don't know by how much... it might be tolerable and still faster than regex spaghetti [00:29]
j_wrightwhy that parser specifically? [00:29]
pharveyCPAN:Marpa
because SvenDowideit said so
[00:29]
gac410yeah. I found it ... ty [00:30]
SvenDowideitgiggle
true tho, i said 'someone _must_ try this out'
[00:30]
pharveybut I like the API. And importantly, it seems to handle multiple tokens starting and ending in overlapping places in the input
which would be necessary for wiki markup generally
[00:31]
j_wrightthe pure perl version is apparently alpha [00:31]
SvenDowideityup
darnit, i need a 1.1.4 dev checkout with unit tests :/
[00:31]
pharveythe spammers haven't spammed my wiki since Friday, after hammering us with 50 registrations per day for 2 weeks :)
er, 20-50.
[00:33]
SvenDowideitgood job to all!
so gac410 pharvey Babar anyone
how do you checkout 1.1.4?
ah, the tag is still useful
[00:33]
gac410gac410 has a directory of tarball installs of each release. [00:34]
SvenDowideitsweet [00:34]
gac410well. some releases. [00:34]
pharveysadly, the repo-per-modules don't have tags yet [00:34]
SvenDowideity, i have that on http://github.com/SvenDowideit/foswiki_releases
but i want a dev version of 1.1.4
so the svn tag is it
[00:34]
gac410Ah. [00:35]
SvenDowideithehe, local git branch of an svn tag [00:37]
gac410pharvey - did you install the trunk email filter patches, or AWSPI? [00:37]
pharveygac410: not yet, I am still working updating to latest prod, but trunk still has issues I'm slowly working towards working on [00:38]
SvenDowideiter mmm [00:38]
pharveyit wasn't too hard to git revert spammer mess
when CDot does GitStoreContrib it'll be even easier :P
[00:38]
SvenDowideitso, adding a translator test, 1.1.4 does the same thing as 1.1.5 [00:39]
gac410It definitely doesn't fail in a live test on 1.1.4 [00:40]
SvenDowideitlets see [00:41]
gac410It isn't pretty - converts the table into ... | A | B <p> </p> <p>C</p> | D | loosing the Table markup. But the newlines survive. [00:42]
SvenDowideithttp://foswiki.org/Sandbox/TaskItem11312
snigger
hhs
pharvey, you plick
i'm trying to make a test
[00:42]
pharveyCancelled :P [00:43]
SvenDowideitwiw
wow
so just going _to_ wysiwyg CD are outside the table
thats toast
you don't even need to save to get a stuffup
[00:43]
gac410You can test on 1.0.9 http://release01x00.foswiki.org/bin/view/Sandbox/TaskItem11312 (or something like that) [00:44]
pharveyyeah, it's in the TML2HTML. [00:44]
gac410yes. indeed. [00:44]
SvenDowideitok poking a test there!
stuffed if i know how of course, cdot writes so little docco :p
[00:44]
pharveySvenDowideit: can you push your branch with the test somewhere [00:45]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/wbUP7Q
[foswiki/Release01x01] Item11312: TinyMCE Corrupting HTML Tables - same as 1.1.4 - SvenDowideit
[00:45]
***GithubBot has left [00:45]
SvenDowideiti did [00:45]
pharveynifty [00:45]
gac410yeah. I started into TranslatorTests and rapidly hit that conclusion. [00:45]
SvenDowideitand updated to show death
dcommitin
[00:46]
pharveythe thing about 1.0.x is that it runs TMCE without forced_root_blocks : true, which is a glorified way of saying that the newer TMCE that we use in 1.1.x *must* have appropriate containers for all textNodes, that means all text must be in a div or <p> or so on... this complicated some of the handling of text around tables/lists, anyway, just sayin.. [00:46]
SvenDowideitnow, lets see what happend to it on 1.1.4
mmm
no
[00:47]
gac410Unfortunately I've never set up a test image for older versions of 1.1 [00:47]
SvenDowideit1.1.4 fails the same way
gac410, its easy
git checkout remote/tags/Release012341234123412
and you can branch it too
and my github repo will give you whats in the zips foreach release
[00:47]
pharveybtw tests run faster if you use PlainFileStore :) [00:48]
SvenDowideitlol
they run faster with mongo too
shame store2 is now very very out of date
[00:48]
pharveyI know :( [00:49]
SvenDowideitpharvey, 1.1.4 has the same test result as 1.1.5
so there's something weirder going on
[00:49]
pharveyok, so I feel as if this is an older artefact of the forced_root_block changes. [00:49]
SvenDowideitso i guess 1.1.4 had code in js that fixed / covered up the problem? [00:50]
pharveyI think your test is failing for the wrong reason [00:52]
SvenDowideitmakes sense
'| A | B C | D |', '| A | B C | D |\x{a}'
yup
[00:52]
pharveyremoving the extra space makes the test pass. Hang on. Let me re-examine what's really happening. [00:53]
SvenDowideitno
its the wrong test
the reported bug shows when you change it to TML2HTML
and there it really is ugly
+ �
there's a foswikiDeleteMe bit in there
so i don't know anywhere near enough
[00:53]
pharveythese were MTempest's tricks to make TML newlines survive HTML transition
they're creating invalid table markup though
[00:57]
SvenDowideitok, and now they don't [00:57]
gac410So if you *create* that same table using TMCE, the cell with the blank line is encoded <p>B</p> <p>&#160;</p> <p>C</p> [00:58]
pharveycrapsticks
<td>B</p>
TML2HTML is probably terminating the table at that bogus </p>
er, sorry. The browser's dom is terminating it there
[00:58]
gac410Ah... that's exactly what SaveWikiPlugin does. An extraneous close will close all enclosing tags. [01:00]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/EoJJIw
[foswiki/Release01x01] Item11312: TinyMCE Corrupting HTML Tables - ah, its tml2html - SvenDowideit
[01:00]
***GithubBot has left [01:00]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [01:00]
pharveyI don't remember the whitespace madness polluting the TML2HTML this badly
pharvey wonders if it was like this before CDot added indedented paras
pharvey checks
[01:00]
gac410pharvey - where do you see <td>B</p> [01:01]
pharveyI copy the HTML into a topic in raw edit, save, and then look at the TML2HTML HTTP response [01:01]
SvenDowideitwish the unit tests worked that way :/ [01:02]
gac410Ah... I was trying that earlier and TML2THML POST was returning an empty response. [01:02]
pharveyit may appear empty because firebug sucks sometimes
or, it was a cache hit in the browser
make sure you're shift-reloading or updating the t=234 param
okay, it was like this before CDot added the paras feature
[01:03]
gac410Must be firebug - updated t= still empty. :(
Over to wireshark
[01:04]
pharveySvenDowideit: I think our test needs to validate to particular HTML - just because it survives roundtrip, doesn't meant the HTML was well-formed [01:04]
SvenDowideitpharvey, y - i changed it to TML2HTML
and that shows something more similar to what you're talking about
but not the same
[01:05]
pharveythe + � appears to be some orphaned $MAGIC_TOKEN [01:07]
SvenDowideitoh.
SvenDowideit mumbles about friggen hashers
pharvey, you need to be careful of pagecache
every variation of every request will put a file into _one_ dir
[01:15]
gac410The <p> </p> rewriting into </p><p><span ... hidden whitespace></span></td> is really broken. [01:20]
pharveyI feel as if MTempest may have fixed lists but didn't quite get to tables.
but that's just heresy atm
[01:22]
***SvenDowideit sets mode: +ooo AlexisHazell gac410 pharvey [01:23]
pharveygac410, SvenDowideit - committed a test which is still perhaps wrong, but should exhibit the problem properly now
sorry, I mean I changed Sven's test
[01:30]
SvenDowideitsweet :) [01:30]
pharveyoooh, I think I found a single line of dodgy code [01:33]
gac410yup ... you have indeed captured it with the incorrect </p> [01:33]
pharveyFoswiki: home of the 800 line method! [01:35]
FoswikiBothttp://foswiki.org/ home [ WebHome ] [01:36]
AlexisHazelli've been browsing around the docs, but can't seem to find any info about how to change the presentation of text in TinyMCE? i want to change how H2+ headings are displayed so that they are the same colour as they are when a topic is 'view'd. [01:39]
SvenDowideityeah, me too :/
i want the styling in mce to be the same as the site styling
but no-one's ever done it for me
[01:40]
pharveyHi AlexisHazell feel free to hijack Item9662 - that's a general issue we should probably fix for everyone [01:41]
FoswikiBothttp://foswiki.org/Tasks/Item9662 [ Item9662: Table borders look odd when you first insert a table using TinyMCE ] [01:41]
pharveythat task says which css file to hack [01:41]
SvenDowideitoh, really? [01:41]
pharveyalthough I think we may have moved where wysiwyg_src lives [01:41]
SvenDowideitit needs to be magic, not a single css file
as i use 3 different skins for different users now
[01:41]
pharveyyou know, Item11312 is not going to be fun [01:43]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [01:43]
pharveythe bug is that TML2HTML doesn't properly understand HTML [01:43]
SvenDowideitpharvey, its already not been fun :/ [01:43]
pharveyit's opening a paragraph just before C and then waits for the only signal it knows of to close it again: another newline
it quietly ignores </td> </tr> etc.
[01:44]
SvenDowideitso the q is, what hack made it ok on 1.1.4 [01:44]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/UG0iUw
[foswiki/Release01x01] Item11312: The test is html-in-tml -> HTML - PaulHarvey
[01:45]
***GithubBot has left [01:45]
SvenDowideitoh shit. home.org.au was working last night, and now its busted again [01:45]
pharvey1.1.4 didn't try to preserve whitespace and newlines. [01:46]
SvenDowideitmmm
ok, so you test fails n 1.1.4 too
[01:47]
AlexisHazellpharvey: Changed pub/System/TinyMCEPlugin/wysiwyg_src.css, that did the trick - thanks. :-) [01:47]
gac410no... Sven, TMCE did the preserving if I understood it. [01:47]
SvenDowideiti just ran pharvey's test on 1.1.4 - it fails [01:48]
pharveySvenDowideit: I expect it would; the 'expected' HTML is just my guess. I'm not sure what it's supposed to really look like exactly. [01:48]
SvenDowideitok, so i can update the tet to pass on 1.1.4? [01:48]
pharveymy expected HTML also has some whitespace protection junk which won't happen in 1.1.4
sure
[01:48]
SvenDowideitor is it all so totally different [01:48]
pharveynah, just fix it [01:48]
SvenDowideityou won't like it [01:49]
gac410er. wotn' tit fail on 115 then because the hiddenwhitespace stuff won't be ther? [01:49]
SvenDowideit<td>B
</p>
<p>
C
</td>
is what 1.1.4 spits out
[01:49]
pharveyand I guess the web browser is quietly coping with it
but not the mess we get from 1.1.5
[01:49]
SvenDowideitso 1.1.4 is backards [01:49]
pharveyI have a theory on how I can fix the 1.1.5 code - but it's a bit scary. This will require ... testing ... and perhaps delaying 1.1.5. It's not a simple fix. [01:50]
SvenDowideithey. [01:50]
gac410Didn't the change to TMCE on 1.1.5 require the special preserving that HTML2TML does now? [01:50]
SvenDowideit<td>B
</p>
<p><span style="{encoded:'s12'}" class="WYSIWYG_HIDDENWHITESPACE"> </span>C<span style="{encoded:'ns9'}" class="WYSIWYG_HIDDENWHITESPACE"> </span></td>
so 1.1.5 just adds lots of spans
but both 1.1.4 and 1.1.5 have the close p then an open p
[01:50]
pharveyit also opens a paragraph tag and never closes it
oh, I see what you mean. 1.1.4 did that too.
[01:51]
gac410pharvey. Based upon what we are seeing I think we have to have you go ahead with the fix [01:51]
pharveythe problem is my fix is more of a hack... :) [01:51]
gac410It's either that or revert [01:51]
pharveyI feel we should probably consider the revert route - but I'll try to get all the tests passing again and commit, then see what cdot says. [01:52]
AlexisHazellOkay, something seriously weird has suddenly started happening with my production Foswiki install. [01:53]
gac410The revert is complicated as well, as I assume TMCE needs to roll back along with MTempest's work. [01:53]
AlexisHazellFirst most of the buttons for TinyMCE stopped being displayed.
Now selecting 'Edit' for a page is taking me to the page for directly editing wiki text.
[01:53]
gac410Hm. Can you undo your last change AlexisHazell
Or is it a file ownership issue - troubles serving the javascript
[01:54]
AlexisHazell(And the "TinyMCE buttons not displaying thing" happened before i messed around with wysiwyg_src.css.)
gac410: Hm, good point, will check that now.
i can only assume our hosting provider has done something, because it was all working .... and now it's not.
Permissions on the files i just changed are fine.
Would the 'edit' script automatically display the non-wysiwyg version if JS wasn't available?
[01:54]
gac410yes I think it does have some fallback. use Firebug to see if any components are not loading
pharvey. Let's plan on an RC2 ... there have been enough changes beyond this that it is probably worth it.
In the scheme of things I'd rather have a really reliable 1.1.5 so we can concentrate on 1.2
[01:58]
AlexisHazell"Uncaught ReferenceError: foswiki in not defined" [02:01]
pharveyAlexisHazell: do you have a customized TINYMCEPLUGIN_INIT(|_TOPIC)? or are you using SafeWikiPlugin? What version of TinyMCEPlugin are you using? Are you seeing 404 errors when the browser tries to fetch the JS/CSS/PNG files from TMCE? [02:01]
AlexisHazellNo customisation of TINYMCEPLUGIN_* [02:02]
pharveygac410: my 'fix' is to HTML2TML first, and then do TML2HTML on that... rather than TML2HTML on manually-entered HTML markup. [02:02]
AlexisHazellNot using SafeWikiPlugin.
This is on a 1.1.4 install.
[02:02]
gac410You have not installed any updated plugins? [02:02]
AlexisHazellThe errors i'm seeing are that the .js files are being retrieved, but that the 'foswiki' var is not defined. [02:03]
gac410pharvey - I'll trust you. I am totally lost within TMCE / WYSIWYG [02:03]
AlexisHazellThough foswiki.js seems to be loading without error. [02:03]
pharveyI think 'foswiki' is established by JQueryPlugin [02:04]
AlexisHazelli just backed out the change i made to wysiwyg_src.css, made no difference. [02:06]
pharveydid the breakage coincide with anything? [02:07]
AlexisHazellUnfortunately not. [02:07]
gac410AlexisHazell: in bin/configure. Extensions JQueryPlugin tab. {JQueryVersion} set to 1.7.1, {JQueryTheme} is Foswiki, and {JQueryPlugin}{Plugins}{Foswiki}{Enabled} is enabled? [02:07]
pharveytried clearing browser cache? [02:07]
AlexisHazellSeriously, one moment the WYSIWYG editor view was working fine; the next moment most TinyMCE buttons wouldn't load; and then finally i lost WYSIWYG editing altogether.
Okay, will do that now.
[02:08]
pharveyalso it's worth restarting your browser.. [02:09]
AlexisHazellgac410: Will confirm that shortly, but i've not been messing around there in configure ....
pharvey: Yes, doing that now - thought it might be easiest. :-)
Browser restarted, didn't fix the issue.
Just checking configure now.
[02:09]
pharveyDo you have firebug? Are there any broken/404 http requests in the net panel? [02:13]
gac410AlexisHazell: This is really strange - [02:13]
AlexisHazellpharvey: i'm running Chromium, using the Developer Tools. [02:14]
pharveywell, in chromium I think it's under the dev tools' "resources" tab I guess [02:14]
AlexisHazellgac410: JQueryVersion is set to jquery-1.7.1, JQueryTheme is foswiki.
gac410: JQueryPlugin is enabled.
[02:14]
pharveyAlexisHazell: if all js is being fetched okay, then I'd start looking at your HTML - perhaps the .js files are being loaded in the incorrect order, for some reason. [02:16]
gac410AlexisHazell: Is it any page, or one particular topic? [02:17]
AlexisHazellgac410: Any page.
pharvey: Okay, i can see that certain GET requests are failing.
[02:17]
gac410That will do it :) [02:18]
AlexisHazellWhy that's suddenly started happening is another kettle of fish .... [02:18]
gac410If it's /pub path then it's apache, not foswiki
If its bin/<script> ... then foswiki
[02:18]
AlexisHazellIt's pub. [02:20]
gac410so something happened either in the apache config, .htaccess files, or file permissions [02:20]
pharveyhah! accidentally nuked Item11312. Note to self: don't use <!-- $WYSIWYG_SECRET --> in literal text :) [02:20]
AlexisHazellFile permissions are certainly fine. [02:20]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [02:20]
AlexisHazelli've not touched the .htaccess files for several days.
So maybe our hosting provider has changed the config of the Apache instance they provide us then ....
[02:21]
gac410what is the fail code? [02:21]
AlexisHazell404. [02:21]
gac410not found then. You can simplyfy debugging by just browsing to that pub url directly - cut foswiki etc. out of the picture [02:22]
AlexisHazellWell, i just removed some Rewrite* stuff from my .htaccess that was previously working - that fixes the issue.
But now a whole lot of confidential attachments are exposed. :-P
pharvey, gac410: In any case, thank you both for all your help!
[02:23]
gac410AlexisHazell: Take a look at the **experimental** Foswiki:Support/NewApacheConfigGenerator - It can generate .htaccess files., i would not use all of them in it's current condition, but it might give you some ideas for the pub rewrite rules. [02:28]
FoswikiBothttp://foswiki.org/Support/NewApacheConfigGenerator [ NewApacheConfigGenerator ] [02:28]
SvenDowideitiment to ask
gac410, in NACG(ed)
you turn off etags
[02:29]
AlexisHazellgac410: Thanks, will do! [02:29]
SvenDowideiti still don't know if i should have them on or off :/ [02:29]
gac410Yes. The recommendaion from yslow was if you use caching correctly the etags just add overhead that is unused. [02:29]
SvenDowideitbut i thought on was more useful
ah, ok
so they make a difference only when the req is made
[02:30]
pharveyin my experience the presence of etags also seem to cause a lot of 304 not modified HTTP requests [02:30]
SvenDowideity, if you don't also have a longer expires
which i do :)
[02:30]
gac410Right. It avoids the full attachment, but adds overhead to every request. [02:31]
SvenDowideitso i get 304's when without i'd get 200's [02:31]
gac410Right. [02:31]
SvenDowideitmmm, it seems not to
in that my cache +11
[02:31]
pharveyI mean, even with expires headers set... but possibly I just had a mangled apache that day. [02:31]
SvenDowideitmeans no req's made [02:31]
gac410Sorrey. Expires. Not cache. [02:31]
SvenDowideitbut when they are made (once every 11 days
they're faster
as they don't contain content
pharvey, its possible that is what happens
[02:31]
gac410Right. The recommendation from yahoo was probaqbly a bit weak. I followed it. But YMMV. I don't recall if they've changed the recommendation in later versions. [02:32]
SvenDowideitme neither :)
i just wondered because there was no comment (and of course, i want to know the why)
what a pain, i have zero motivation today
guess i need to go set up my bike
[02:32]
gac410When browser has cache, etag says "verify if it has changed" expires says, don't even check. It adds a small amount of overhead to every get, making every request bigger. [02:33]
SvenDowideitnot quite
it makes every _full_ 200 get a tiny amount bigger
[02:34]
gac410Yes. [02:34]
SvenDowideitbut it also means that when the expirely happens, it would be possible to juet get a 304 - no content [02:34]
gac410gac410 wonders if 304 resets the cache expires. [02:35]
SvenDowideitso 99% of etag post expirely responses are much smaller, as they are just a 304
ah - thats where i _hope_ :)
[02:35]
gac410right. I am not really arguing either way. I was following the advice from "yslow" that was giving my site a "B- Grade" with recommendation to dump the etags [02:36]
pharveyIIRC you need to use etags with Cache-Control, or Expires, but not all three... or something.
pharvey wonders where the best place to add his hack should be >:(
I feel as if it should be in the rest handler, but then unit tests will miss out.
[02:37]
SvenDowideitpharvey, where its fast&obvious
and then we force CDot to poke it too
[02:39]
pharveyI'm also hard-coding some options on the HTML2TML conversion, as our existing API doesn't really cope for this additional step :(
oh man. This breaks every other test :)
this approach isn't going to work. Damnit
[02:42]
gac410IIRC CDot did look at the task - saw "newline eating" and said he would not touch it with a <some length> pole [02:47]
pharveylol
so, to be clear
this bug affects people WYSIWYG'ing topics containing HTML
and they always got mal-formed HTML in the past
it's just that our whitespace-preservation markup is making that even more sad
'just'
[02:48]
gac410Yeah. the simple solution is to add html to the default for skipping wysiwyg. WYSIWYG_EXCLUDE ... it does work. [02:49]
pharveymaking the whitespace preservation configurable would be a cop-out wouldn't it.
The WYSIWYG code really isn't designed for mixed TML/HTML
not complex HTML like lists and tables at least
[02:49]
SvenDowideiti'd rather we started with exclude [02:50]
gac410Maybe for 1.1.5, we just add html to the default exclude list. [02:50]
SvenDowideitand then slowly increased the number of things we don't exclude
especially as the guys i upgraded last month
[02:50]
gac410The exclude is for "html" [02:51]
SvenDowideituse alot of html for presentation [02:51]
pharveyyou're always going to have some html in plain old TML [02:51]
gac410I've been unable to make explicit exclude for <table> working. [02:51]
pharvey<img for a start [02:51]
SvenDowideitgac410, i added code for script and style tags [02:51]
gac410The "html" exclude does have exceptions [02:51]
SvenDowideityou can copy that for table [02:51]
pharveyso, I desperately have to get back to real $work, sorry I couldn't fix it :( [02:52]
SvenDowideitsorry to take your morning :( [02:52]
gac410okay. Thanks for the try pharvey. We've still learned a lot [02:52]
pharvey'tis what I shoulda done on the weekend :P [02:52]
SvenDowideity, but reality is what it is [02:53]
gac410Okay I have a simple fix. Added WYSIWYG_EXCLUDE = script,style,table (as a new default)... and code to exclude wysiwyg if html table markup exists. [03:00]
SvenDowideitSvenDowideit likes it to a point
what happens if you thn create a table using wysiwyg
oh, that'll still work ok
until ..
[03:03]
gac410I'll test. It *should* be just fine.
As long as it saves as TML, it will still be fine.
It's draconian, but I think it's safest for 1.1.5. either that or we revert and loose the new TMCE editor
[03:03]
SvenDowideiti agree [03:07]
gac410foo. The WysiwygPlugin.txt file changelog - has not been maintained. And the plugin has had some "Sync from trunk" type commits. [03:14]
SvenDowideiti thought about rejecting those
cos they're written by dumb people
[03:15]
gac410Unfortunately it takes forever to try to backtrack which tasks were addressed, and then re-target them to 1.1.5, and update changelog. and often fix release notes. Royal PITA [03:17]
SvenDowideity, i was surprised you didn't tell whoever did the merge to revert or do it
then again, i was surprised we have the trunk one in 1.1.5 period
SvenDowideit is clearly in two minds about most things
[03:17]
gac410It's been never ending battle. JQuery has been a nightmare. And then TMCE now WYSIWYG [03:18]
SvenDowideitso what we need to do
is to work out what code we can write to automate more of the nightmare
maybe we just need to insist that code that goes into non-trunk == cherry-pick or rigourously marked
[03:18]
gac410Sometimes "Sync from trunk" is because the entire module(s) were copied. Other times, it's a cherry-pick of a commit with a poor comment.
Can't tell without trying to compare
Deja Vu all over again. .... didn't I have this discussion with you a few weeks ago?
[03:20]
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/YZoU0Q
[foswiki/Release01x01] Item11312: Also WYSIWYG_EXCLUDE html table - GeorgeClark
[03:32]
***GithubBot has left [03:32]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [03:32]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/b0rwpA
[foswiki/master] Item11713: Fixed. - AlexisHazell
[foswiki/master] Item11714: Initial checkin. - AlexisHazell
[03:33]
***GithubBot has left [03:33]
FoswikiBothttp://foswiki.org/Tasks/Item11713 [ Item11713: AcronymDefinitionsPlugin breaks autolinking when web name is a defined acronym ]
http://foswiki.org/Tasks/Item11714 [ Item11714: Create FSBrowserPlugin ]
[03:33]
SvenDowideitgac410, yup
main thing is we need to make the TM job easier
RM
cos its too hard and takes too much&too long
[03:34]
gac410I'm going to cherry-pick all the fixes into trunk, and will probably set the test to 'expect failure' until we get a real fix. [03:35]
:P Trunk now needs File::Remove? [03:44]
GithubBot[foswiki] foswiki pushed 5 new commits to master: http://git.io/H8F9xQ
[foswiki/master] Item11312: TinyMCE Corrupting HTML Tables - same as 1.1.4 - GeorgeClark
[foswiki/master] Item11312: TinyMCE Corrupting HTML Tables - ah, its tml2html - GeorgeClark
[foswiki/master] Item11312: The test is html-in-tml -> HTML - GeorgeClark
[03:46]
***GithubBot has left [03:46]
gac410Disabling wysiwyg for table breaks 3 tests, since they are now excluded :( Somehow need to override WYSIWYG_EXCLUDE in that test somewhere :P
Sometime tomorrow - I',m falling asleep here. g'night all
[03:53]
SvenDowideitnite [03:55]
***gac410 has left [03:55]
pharveygac410, SvenDowideit - it was a "sync from trunk" commit probably because I convinced CDot that trunk was stable and good, as we've been using trunk WYSIWYG on trin.org.au continuously for about 2 years now, combined with the fact that between Wysiwyg & TMCE there have been 5 different authors committing work to trunk and none to Release01x01 which became stale [03:56]
SvenDowideitpharvey, so? - when i'm hennypenny i see patch :)
i marvel that trunk had something for 2 years that the stable release didn't
clearly, we should get more sensible about this
next major release.... june
[03:58]
pharveymy "2 years" story doesn't mean we never did any work on Release01x01 for 2 years, just that this is what I've been using in prod [03:59]
SvenDowideitno, it means that the 1.1. version of mce is ancient
even compared to trunk
which i presume is ancient compared to mce head
[04:01]
pharveywe're using a november 2011 release [04:02]
SvenDowideitin dev
but in 1.1.4?
SvenDowideit knows nothing :) just thinks all this is a problem specifically because we're too chicken to release a major once or twice a year
[04:02]
pharveyTraditionally my routine for doing a new TinyMCE upgrade involves about 2 hours of browser testing in IE6,7,8,9 + Firefox + Chrome + Safari
I never found that 2 hours to do all the browser testing
[04:03]
SvenDowideitergle [04:04]
pharveyI've been burnt in the past by TMCE's IE "fixes" that actually makes things worse (okay, it was only once, the 3.3.9.4 IIRC)
the lack of frequency of TMCE updates reflects the amount of work required to do all the testing.
and the lack of our ability to do it.
or mine, at least.
which is why I need to improve the selenium setup (and in this case, at least fix them).
[04:04]
SvenDowideitSvenDowideit is frightend [04:07]
....... (idle for 31mn)
argh.
foswiki.fcgi and view both work from the cmdline
but it doesn't from apache
| 2012-04-01T05:54:10Z debug | SEARCH crash: System.WebChanges: SEARCH{"'1'"...
Caused the following error:
Illegal division by zero
[04:38]
AlexisHazellAs distinct from a legal division by zero? :-) [04:49]
SvenDowideitgiggle [04:49]
AlexisHazell, FSBrowserPlugin - awesome
i've a pile of data dirs i might use that on
[04:58]
AlexisHazellSvenDowideit: Ah good! :-) [04:59]
SvenDowideit100.43.83.153 - - [01/Apr/2012:21:59:59 -0700] "GET /bin/login/Homeorgau/WebHome?t=1269958678;foswiki_origin=GET%2cedit%2c/bin/edit/Homeorgau/WebHome%3ft%3d1269958678 HTTP/1.1" 200 2393 "-" "Mozilla/5.0 (compatible; YandexBot/3.0; +http://yandex.com/bots)"
gawd bots are dumb
mmm
i thought we changes login to return 401 :(
mmm, pharvey dumb q
do you think that its worth letting bots add query params to req's at all
i could remove them :)
meh, that's icing for next month
[05:00]
i think i've got it.
sadly 99% of the bot requests are for idiot url_params
so i think i will kick them to the curb
hehe
looks like i accidentally am
[05:15]
......... (idle for 40mn)
BabarSvenDowideit: git svn fetch should fetch you the tag, and then you just checkout FoswikiRelease01x01x04 [05:57]
SvenDowideitBabar, yup
thats what i did :)
argh!
[05:58]
AlexisHazellIf i don't want to use /either/ gmtime or servertime, is there some way of specifying a third option? [05:58]
SvenDowideitso we use BrowserMatchNoCase
for blocking bots
except its useless
i can't recal atm :/
i did something to do with localtime dislpay in a plugin once
used the TZ setting on a user's topic
but its been a long time
[05:58]
AlexisHazellHm. [05:59]
SvenDowideit: http://foswiki.org/Tasks/Item10039
SvenDowideit: Seems to me that maybe the best thing for me to do might be to develop a TimezonePlugin ....
[06:07]
SvenDowideitsimpler to use one of the ones that exist [06:11]
Babarwhat are you trying to do exactly/ [06:12]
SvenDowideithttp://foswiki.org/Extensions/LocalTimePlugin
is the one i extended
make DISPLAYTIME be useful i bet
[06:13]
AlexisHazellSvenDowideit: Yes, i looked at those, but i want the correct time displayed for last edited time etc. [06:14]
SvenDowideitAlexisHazell, one trick i can't recal
SetEnv TZ Australia/Brisbane
in my apache conf
[06:14]
Babaryeah... it's not possible to configure the apache server properly? [06:14]
AlexisHazellThat wouldn't work via .htaccess would it? [06:14]
SvenDowideittry it [06:14]
Babarit might :) [06:14]
SvenDowideiti have that in my dreamhost htaccess file
i put it there, um, >5 years ago
right now i'm hacking pagecache to serve bots from static html
using mostly htaccess
[06:14]
AlexisHazellNope, SetEnv in .htaccess didn't work.
So my thought was
somehow pass dates+times through a mangling process involving the DateTime module.
[06:17]
SvenDowideitfirst thing - does that patch not work for you?
cos imo it would make alot of sense to have such a feature
we just need someone to take it and make it releaseable
[06:21]
AlexisHazellSvenDowideit: i'd not tried the patch; my first thought was that maybe it might be better to test out solutions that don't require a patch. But i'm happy to try giving it a go; i imagine it will need some finessing to work with the current code. [06:24]
SvenDowideiti think you have no choice
if you can't tell aapche to setyour env
(which your hoster might have done)
as you want DISPLAYTIME to do somethign useful
[06:24]
AlexisHazell*nod*
Does core already have DateTime as a dependency?
Because i'd rather use that than insert a whole lot of custom timezone manipulation code into core.
[06:25]
Babargreat. core Time handling wasn't that complicated already... [06:27]
SvenDowideityou need to see the date proposal that arthur made
i can't recal
Babar, its not complicated
its handcoded shit
[06:27]
Babarnah, for the moment, everything is done manually [06:27]
SvenDowideittakes 12 times as much code to do 20 times less [06:27]
BabarSvenDowideit: yeah right. And unit testing with different settings isn't complex either :p [06:28]
AlexisHazellSvenDowideit: Well, i'm looking at Item10039 right now, which i presume is the thing of George's you're talking about? [06:28]
FoswikiBothttp://foswiki.org/Tasks/Item10039 [ Item10039: add possobility to redefine timezone in user settings ] [06:28]
Babarcan you fix the typo then please? :) [06:28]
SvenDowideitya [06:28]
AlexisHazell(Typo in item summary fixed.) [06:30]
Babarthanks
and I don't understand Lavr's comment about "Patch assumes only full hours"
[06:30]
AlexisHazellThe issues raised in the discussion for that item would partly be addressed by using DateTime. [06:31]
Babarhas this been fixed, or... [06:31]
AlexisHazellBabar: Not all timezones differ by a full hour.
Adelaide, for example, is 30 minutes behind Melbourne.
[06:31]
BabarAlexisHazell: thank you, I'm perfectly aware of that [06:31]
AlexisHazellWell, my apologies for not understanding what you meant by not understanding Lavr's comment. :-) [06:32]
BabarBabar doesn't see in what way the patch assumes only full hours
and the _tzOffset is a mess I've had to fix. Several times.
took me a long time to get it right, and get the unit tests proper too
which is why I say that: 1. it's complex, and 2. it's hard to unit test
and I think the patch it totally flawed btw :)
[06:32]
AlexisHazellWell, i'll need to be working on this in any case, because it's going to be a significant issue for my users.
So i might just have to mess around with our install to get some sort of workaround.
[06:35]
Babarthen you'll need to test BOTH ways. [06:36]
AlexisHazell"both ways"? [06:36]
Babarwhat I mean is that people often test that the times are displayed properly, and that's it
but... users enter times too. And that needs to be saved as the proper epoch, so diffed with the proper offset between the server's timezone, and the user's
[06:36]
SvenDowideity, and right now, foswiki is terrible [06:37]
Babaragreed [06:37]
SvenDowideitas it uses some tz that the user is not in [06:37]
Babarbut it works©® [06:37]
SvenDowideitso its going to be hard to make it worse
no, it does not
it might be mathematically correct, but its disfunctional
[06:37]
Babarit does not? Show me a bug report :p [06:38]
SvenDowideitthat one AlexisHazell has is a perfect eg of how completely stuffed it is
its right in the mathematically symetrical sense
[06:38]
Babarnah. it has a major flaw: it gives people way too many freedom
and that's a general issue
[06:39]
SvenDowideitbut totally useless for human beings that don't do tz calcs in their head and know the server's tz
do you know the tz of my server?
[06:39]
AlexisHazellWell, i guess another option is to lodge a ticket with our hosting provider to see if they're willing to do a SetEnv TZ for our VHOST. [06:39]
SvenDowideitcos i sure as shit don't [06:39]
BabarI can. %GMTIME% vs %SERVERTIME% [06:39]
SvenDowideitAlexisHazell, i would try that tho :)
yes, you _can_
but you don't actually know it
[06:39]
Babarnope. I'm assuming the admin sets a reasonnable default [06:40]
SvenDowideityou'd have to look it up and then figure out some maths to enter a time that then works right
its reasobale
for the physical location of the hoster
[06:40]
BabarI agree it's far from perfect [06:40]
SvenDowideitthats the understatement [06:41]
Babarbut at least, once I know it, I can work around it [06:41]
SvenDowideitits broken, and consistently so [06:41]
Babarif I try to make it clever, I might end up shooting the users in the feet [06:41]
SvenDowideitas most people can't work around thigns like that consistently [06:41]
Babarfor me, the server TZ should be irrelevant [06:41]
SvenDowideity [06:41]
Babarit should have some preferences that defines how it's displayed [06:42]
SvenDowideittahts why i always want to talk about DISPLAYTIME and DISPLAYNAME and stuff
thats what LocalTimePlugin did
but the interest was too minimal for me to continue
[06:42]
Babarthen I think the proposal for the task above is: resurrect LocalTimePlugin and make it core :)
that would be a killer feature... especially for my users, half of them being on the other side of the pond :)
[06:42]
SvenDowideitthe patch is i think a better impl
or a combination of things
meh, still battling htaccess
[06:43]
***ChanServ sets mode: +o MichaelDaum [06:48]
AlexisHazellWell, i've lodged a ticket with our hosting provider, will see what they say. [06:57]
SvenDowideitcool (i knowits mean, but i hope they encourage you to implement the feature
well.
I now have a foswiki that serves static html from the cache with zero perl code
and for good measure the bot i'm watching 's requests work, but any with url params get 301'd back tot he non-url param version
now i have the problem that web does not exist returns 200
oh.
MichaelDaum, pageCache is caching oopes
so 'web does not exist' becomes shithotfast
[06:58]
AlexisHazellSvenDowideit: *chuckle* Well, if i have to take it on, sobeit; but if i can avoid it for now .... :-) [07:02]
SvenDowideityup
hehe
MichaelDaum,
[07:02]
BabarBabar would advice you not to touch with even with a 10 feet pole [07:02]
SvenDowideitpharvey, CDot and anyone that wants to see pagecache fast [07:02]
Babarbeen there, done that. [07:02]
SvenDowideitab -n 24 -c 12 http://uwiki.net/view/Homeorgau/SmallProjects
min mean[+/-sd] median max
Total: 351 382 48.3 372 541
Complete requests: 24
oh, and
Waiting: 173 193 41.9 182 332
and this from my home to the DH shared server in the US
[07:02]
BabarTotal: 320 335 45.0 323 482
from work :)
[07:04]
SvenDowideitgrin [07:04]
CDotwtf? How can createNewFoswikiSession result in the modification of Main/SitePreferences and System/DefaultPreference? [07:05]
SvenDowideitfor a foswiki, i think we're doing well
CDot, why would it?
[07:05]
BabarTotal: 364 490 118.8 468 742
from home
[07:05]
CDotyou tell me; i can't work it out [07:05]
SvenDowideiti don't recal seeing it do that [07:05]
Babarmodification? it should access it, not modify it [07:05]
SvenDowideitSvenDowideit gets to run away tho - time to pick up kids [07:05]
CDotme neither, but that's what's happening :-( [07:05]
SvenDowideitits probly a bug in plainfile.... :p
god bots suck
System/VarTEAL?skin=widgets;widgetscolumns=wm;colsidx=6;skin=widgets;widgetscolumns=wtt
back later
pharvey, i think i have some fin stuff for you
[07:06]
CDotaha; sussed it. The PlainFileHandler is quite smart about fixing up topics where the history is inconsistent with the latest. It does a check on every topic read. And those two topics, in svn, are in a broken state :-( [07:08]
Babaryeah... bots are the main reason f.o gets slow from time to time, and the machine gets overloaded [07:08]
CDotso it's fixing the META:TOPICINFO in them
CDot goes back tothe drawing board
[07:09]
SvenDowideiti'll update our botkiller conf this/next week [07:09]
Babar(when Yahoo!, bing, google and baidu index at the same time, poor server suffers) [07:09]
SvenDowideiti'm basically 301 ing all urlparam bot requests to the topic [07:09]
MichaelDaummorning all
SvenDowideit, which page cache is it you are testing? trunk?
[07:10]
AlexisHazellBabar: Yes, i get the impression you've been rather badly burnt by this! [07:11]
BabarAlexisHazell: by what? The Time code? Yeah, I was. [07:11]
AlexisHazellYes, the time code. [07:11]
MichaelDaumnote that ab is not likely to produce any interesting data. it only is testing one url, which is pretty pointless. [07:12]
Babarit's the problem when I start looking into some code, and realise it's wrong... it creates an inch I have to scratch :(
s/inch/itch/
[07:12]
pharveySvenDowideit: we have non-human/non-web-browser agents which need to pass in URL params :/ [07:14]
.... (idle for 17mn)
AlexisHazellOkay, am i wrong in thinking that my hosting provider telling me to alter php.ini is not going to work?
Can Apache in fact draw on php.ini for /any/ CGI scripts it runs?
PHP or not?
[07:31]
Babarnope :) [07:38]
AlexisHazellBabar: Thank you. :-)
i have however gone through the process of specifying date.timezone in ~/php.ini, so that the tech himself can see it doesn't work.
Also, he directed me to a knowledge base article that suggests SetEnv TZ in .htaccess /should/ work. :-P
[07:44]
Babaryeah, you could have mentionned you tried it already :) [07:48]
AlexisHazelli did. :-)
Turned out, though, that SetEnv TZ does work - it was just that i hadn't actually applied my change in configure to use servertime. *blush*
(i.e. had made the change to the relevant field, but hadn't actually saved the change.)
In any case - that saves me having to deal with the Time code.
[07:54]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/6bn5Ng
[foswiki/master] Item11710: added {ExcludeJavaScript} and {ExcludeCss} feature to specify files not to be processed by the asset optimizer - MichaelDaum
[08:03]
***GithubBot has left [08:03]
FoswikiBothttp://foswiki.org/Tasks/Item11710 [ Item11710: POP incompatible with TMCE ] [08:03]
.... (idle for 18mn)
SvenDowideit<pharvey> SvenDowideit: we have non-human/non-web-browser agents which need to pass in URL params :/
so?
they're not in your 'itsabot' detection list
so don't get the bot indexer response
MichaelDaum, course
trunk pagecache plus a new backend that subclasses mysql
and adds static files
[08:21]
MichaelDaumpop? [08:22]
SvenDowideitwhich means i can view the files and see what's weird
pop?
[08:22]
MichaelDaumpage optimizer pluggy [08:23]
SvenDowideitgod no
this is core+homepage+fastcgi
not even what we distribute
super minimal, and really interesting
[08:23]
MichaelDaumah, seen your bug report.
was wondering what you were aiming at tho...
[08:23]
SvenDowideitsimple :)
avoiding burnout
i have zero motivation atm, due to illness and having just delviered the per-cairo twiki conversion
so i thought i'd try pharvey's git via pseudo-install
and leverage pagecache to serve static files via apache to google
[08:23]
MichaelDaumya. well pharvey is our git sensei. [08:25]
SvenDowideitSvenDowideit mumbles oliphant [08:25]
CDotSvenDowideit: we talked a while ago about moving META into the store. Specifically META:TOPICINFO and META:FILEATTACHMENT should not get near the higher levels of core; agreed? [08:25]
MichaelDaumcool to see people play pagecaching :) [08:25]
SvenDowideitits a much more understandable approach than last time [08:26]
MichaelDaumy [08:26]
SvenDowideitthough i need to work through your unit tests
so i know i'm not breaking things
CDot, mmm, except :/
[08:26]
CDotthat's why I'm asking.... [08:27]
SvenDowideitbasically, i think Serialisation should not be _in_ the sotre
store
but it should be used by the store
gotta go
back in a few hours
[08:27]
CDotcos it seems to be Store should popluate and maintain and serialise and etc. FILEATTACHMENT and TOPICINFO [08:27]
MichaelDaumMichaelDaum adds "bulk operations on meta" to the wish list [08:27]
CDotin fact, I'm having major problems ATM because core insists on f**king with them
CDot doesn't care where serialisation is; more concerned with who maintains FILEATTACHMENT. Trying to close the gap between "what I think is in the store" and "what is really in the store"
[08:27]
MichaelDaumCDot, shocking how much code addressing META is inside Foswiki::UI:** ... a place nobody would expect these things to be [08:29]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::UI [08:29]
CDotI don't have a problem with *accessing* META. I have a problem with *writing* it.
cos the %META statements get out of step with what is actually in the store
[08:30]
SvenDowideitbasically [08:31]
CDotCDot has already cracked it for TOPICINFO in PlainFileStore (which maintains TOPICINFO) [08:31]
SvenDowideitserialisation
is used for more than storing
its used for rendering (aka communicating) too
so while the meta contains the parsed data
[08:31]
CDotsure, give me a serialiser and I'm happy to use it. I'm more concerned with how META is maintained [08:32]
SvenDowideitthe serialisation service is used to convert that into json/tml/blah
its there
[08:32]
CDotI want to push all ops on FILEATTACHMENT and TOPICINFO down into the {Store}{Implementation} [08:32]
SvenDowideitthe serialisation classes exist and are in use [08:32]
CDot^all ops^ all change ops [08:32]
SvenDowideitso store becomes the rpm beast that modifies the meta obj?
then er, make a proposal
i don't want to have to totally re-re do store 2
[08:32]
CDotCDot is going to, just wanted to shoot the breeze first [08:33]
SvenDowideitits stalled waiting for you and others to look
as it probably alreadys does half what you're talking about
[08:33]
CDotI'm working towards it. What I have done so far is 100% compatible [08:33]
SvenDowideitand the store2 proposal probabl already covers it to some degree [08:33]
CDotCDot is working in trunk [08:33]
SvenDowideit100% compat is not useful
as compat is shit.
[08:33]
CDotit's a starting point [08:34]
SvenDowideitlet me put it this way
the last 3 times i worked on re-doing store
[08:34]
CDotlast time I looked at store2, there wasn't much to see.... [08:34]
SvenDowideiti had to throw away all my code
ecause you made big changes
this time
i published the code so you can help
[08:34]
CDotif it's in trunk, then I'm using it [08:34]
SvenDowideitthen you didn't look very hard
its not
its in github, like we told you
in that case
please stop playing in store
i've spent too much time fixing your last 2 changes
if you're not going to read what i'm doing
SvenDowideit has to go for dinner*girls
[08:34]
CDotCDot is too sick to argue. Fuck it.
CDot wanders off to do something more interesting instead.
[08:36]
..... (idle for 24mn)
ModAcOstHi, I want to create a personal page for each user with a custon view template. However if I set Vi
VIEW_TEMPLATE every page will have it set
Thats a feature right?
[09:00]
SvenDowideity
use local rather than set
[09:03]
ModAcOstah, didn't know that [09:04]
SvenDowideiti always forget too :) [09:04]
AlexisHazellSvenDowideit: i have a kludgey fix for http://foswiki.org/Tasks/Item11716. It doesn't actually solve the problem at its source, just does some surgery afterwards; and i've taken this approach because i just can't work out how create_new_extension.pl creates the contents of MANIFEST, even after stepping through it a few times with the debugger.
So what's our position on such kludgey fixes?
[09:09]
CDotSvenDowideit: ok, I looked at store 2. It's sufficiently diverged that it's a complete rewrite of the PlainFileStore - which is OK, don't mind that. You didn't move the FILEATTACHMENT management down into the store, so I will add that to the todos for store2, and not bother for store1. [09:12]
BabarCDot: how do you say thanks in Scottish Gaelic? Thank ye?
Moran taing? (the other one was Scotts, sorry)
[09:21]
CDotdepends; are you speaking to one person, or many? [09:21]
Babarone [09:22]
CDot"tapadh leat" is nicely informal [09:23]
Babarthanks :)
Babar hopes he speaks Gaelic... but he's from Glasgow, so he should
[09:23]
CDotdoubt it
Gaelic is for teuchters only. The sassenags don't learn it in school.
[09:23]
........ (idle for 38mn)
SvenDowideit: BTW were you aware that the POD doc extractor built into the core only handles a subset of the POD tags? Using full =head2 style tags, as is done in Store2, won't work :-(
Unless, of course, you also extended the extractor.
[10:02]
Babarand once again, we reinvent the wheel
because we want it to be able to parse TML?
[10:04]
CDoty
it just rips out =begin TML ... =cut
it *could* be rewritten to use CPAN POD, but that would have been more work.
CDot just re-used code that was already there
[10:04]
Babaryeah, it could just clone Pod::Trac, and fix it to understand TML [10:06]
CDotit doesn't do anything with TML; it just extracts comment blocks. So it's not really POD, except for the tag syntax.
I assume the author only used =begin ... =cut because the compiler ignores it
[10:07]
Babarhttps://metacpan.org/module/Pod::Simple::Wiki::Twiki::Upload
oh look.
[10:08]
CDotCDot isn't going to do anything about it; was just observing. [10:08]
Babarhttps://metacpan.org/source/JMCNAMARA/Pod-Simple-Wiki-0.14/lib/Pod/Simple/Wiki/Twiki.pm
Pod::Simpe::Wiki supports Twiki
ok, lunch
[10:08]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/YB5cSw
[foswiki/master] Item11708: documentation corrections, and one code correction to be consistent with doc (zero impact as it is never called in this mode) - CrawfordCurrie
[10:17]
***GithubBot has left [10:17]
FoswikiBothttp://foswiki.org/Tasks/Item11708 [ Item11708: RcsWrap and RcsLite do not implement the documented Store interface ] [10:18]
CDotlordy, that POD module is crude. It doesn't support headings, for e.g.
CDot weighs up the pros/cons of (1) rewriting all TML in POD comments to POD markup (2) writing a complete TML renderer in the CPAN POD module (3) leaving it alone
[10:21]
LavrI am walking through all bugs fixed in 1.1.5 now. I am half way. So far they are pretty well characterized. I am beautifying the headline when I feel it improves the understanding
lots of good fixes in 1.1.5
[10:28]
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/HiWQNw
[foswiki/master] Item11708: add low level tests for the current store API, developed for PlainFileStoreContrib. Note that these tests will be redundant in Store2, but until then they fill a hole. - CrawfordCurrie
[foswiki/master] Item11689: verify against the new StoreImplementationTests - CrawfordCurrie
[10:30]
***GithubBot has left [10:30]
FoswikiBothttp://foswiki.org/Tasks/Item11689 [ Item11689: Release PlainFileStoreContrib ] [10:31]
***chriswerry has left [10:39]
.......... (idle for 47mn)
SvenDowideiti am missing something :/
why do things like piwik exist?
doesn't my apache access log give me that same data without stuffing the user's experience?
(i mean the js based piwik gathering stuff
[11:26]
ColasHomepiwik gives you a way to capture what is going on outside the url (hidden parameters, session-based data, ...)
I guess having a way to dump this info into apache log could bring you best of both worlds
[11:34]
SvenDowideitso its a way to get into the server stuff the server already knows
SvenDowideit giggles at the 'design'
[11:37]
MichaelDaumpiwik's js sometimes is slow -> should be loaded async'ly by injecting the script tag into the head after the document has finished loading.
doesn't it track clicks using cookies?
[11:37]
SvenDowideitthe entire idea of loading js to track something my server SW knows ? - cos y, i don't like them
becuase they do slow things down
y
last time i looked, every html page i generate goes through my server already
and i have access to session cookies etc in my apache conf :)
[11:38]
MichaelDaummy guess is: recording click trails is more reliable mapping it onto a cookie fingerprint [11:39]
SvenDowideiti guess the killer feature is that it 'just works' for people that know little about stuff [11:40]
MichaelDaumthat too [11:40]
SvenDowideitok, resolved :) there no benefit for my use, as i have apache skilz and don't need to slow down the user's browser
next version of piwik will be more useful
1.7.2 has an apache log importer
[11:41]
Babaryeah, but I'm guessing one needs to tweak the apache logs to dump cookies [11:43]
pharveyjQuery.getScript() [11:44]
ColasHome<SvenDowideit> i guess the killer feature is that it 'just works' for people that know little about stuff => that's the main point. [11:44]
SvenDowideitSvenDowideit mumbles
what a waste of billions of cpu watts
[11:46]
Babar"The installation takes only 5 minutes! How do Install Piwik?" is that proper English? [11:46]
SvenDowideits/do/to/ [11:47]
Babars/I/& i/ I'd say [11:47]
ColasHomeAnyways, it "does the job" perfectly. We put piwik on th internal sites we set up for teams, and they use it [11:48]
SvenDowideitit does [11:48]
pharveyI read a job description today, "you are collect excellent communication skills" [11:48]
SvenDowideitbut when you want your pages to be fast all the time
then adding one more flakey sometimes slow for no reason thing to track stuff you can actually configure apache to get you in the first place
then you start to wonder
[11:48]
Babarthe more I look at piwik, the more I like it: http://piwik.org/roadmap/ first link: #703 links to bug #583...
http://dev.piwik.org/trac/ticket/703 <--- is still not done Sven
[11:49]
SvenDowideitI now have a foswiki that sense non-perl based statis html to indexers
Babar, it says you can play with it in 1.7.2 (rc)
[11:49]
ColasHomehowever, that could be a nice plugin for foswiki: advanced analytics [11:50]
Babar"I did quite a few experiments, and eventually found that 40 is about right. This is a VM running on a high powered Dell R710, so although the OS only thinks it has 4 CPUs I don't know how things actually pan out. All I know is that the number of records/sec increases pretty much linearly with --recorders up until 40. E.g. if I run at 32, I get more like 200r/sec rather than 250+r/sec. A single recorder manages around 6-7r/sec. After 40 the ...
... benefits tail off. "
OMFG.
Dell R710 have 2x six cores, usually, so with hyper threading, it sees up to ... 24 "CPUs". I wonder where he got his 4 CPUs from
but hey, it's php and python. Nothing surprises me
[11:50]
SvenDowideitgrin [11:52]
Babar"Probably set up a cron job to send a SIGHUP to apache every hour or so, so it restarts the importer script (in case of memory leaks, etc). "
this bug report is a goldmine :)
[11:52]
SvenDowideitSvenDowideit is close to pointing Babar at some of our old bugs :p [11:53]
Babarnot saying we're better, just making fun of it
hey, come on, today I'm allowed :p
Babar checks which piwik version we have on f.o :)
1.1.1 :(
[11:53]
SvenDowideitwell, i installed it
ran the importer
and it dies
on a pretty simple access.log
[11:55]
Babarfunny. can't seem to find how piwik was installed, on f.o...
funny because I thought *I* installed it :)
[11:56]
SvenDowideitwe moved everything 2 dir's to the right [11:56]
Babarah no, it's 1.7.1 [11:57]
SvenDowideitand then ran a dictionary match on the rot13 of the filenames to make new ones [11:57]
Babarhum... shall I try to upgrade? You said 1.7.2 anyway [11:58]
SvenDowideit1.7.1 is the latest release
and no, so far i'm underwhelmed and will use a log analyser
[12:00]
AlexisHazellCan i be trusted enough to be granted ALLOWTOPICCHANGE on f.o/System? :-) [12:01]
BabarNO! [12:02]
SvenDowideityes :p [12:02]
AlexisHazellAlrighty then. :-) [12:02]
Babargac410 will kill you [12:02]
SvenDowideitand you can write the code i said i would write and then didn't
gac knows i'm a bum :)
[12:02]
Babaras then he will have to merge things back when upgrading... [12:02]
SvenDowideitAlexisHazell, System web is managed in svn
so as you have commit :p
[12:02]
AlexisHazellSvenDowideit: Ah, okay. [12:03]
SvenDowideitand gac410 wil lbe building an rc2 and updating f.o with it (i think) [12:03]
Babaroh right. Stupid nick_color.pl tricked me once again into thinking AlexisHazell's sentence had been uttered by SvenDowideit ... [12:03]
SvenDowideitBabar, :p
perl sux?
[12:03]
Babarbut still, my answer holds. I wouldn't have been that harsh if I would have known it was Alexis :) [12:04]
AlexisHazell*laugh* [12:04]
Babarnah. This algorithm is a bit odd at times [12:04]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/h2Ovhg
[foswiki/master] Item11716: Kludgy fix. - AlexisHazell
[12:17]
***GithubBot has left [12:17]
FoswikiBothttp://foswiki.org/Tasks/Item11716 [ Item11716: BuildContrib's create_new_extension.pl does not add appropriate names to MANIFEST ] [12:18]
...... (idle for 28mn)
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/KV28Iw
[foswiki/master] Item11718: System.AccessControl topic page amended. - AlexisHazell
[12:46]
***GithubBot has left [12:46]
SvenDowideitmmm
ok, MichaelDaum
i've run the page cache for about 12 hours
[12:46]
FoswikiBothttp://foswiki.org/Tasks/Item11718 [ Item11718: System.AccessControl should note {AuthScripts} affects access control. ] [12:46]
SvenDowideitand now my fastcgi's time out lots
i'll debug some other time, as its late here
[12:47]
MichaelDaumk [12:47]
SvenDowideitbut turning off pagecache and bouncing the fcgi processes brings it back
so there's someting mean happening
[12:47]
MichaelDaumwhich DBI are u using? [12:47]
SvenDowideitmysql
uwiki.net gets hit constantly by bots tho
so its a harsh place
[12:47]
MichaelDaumgood test [12:48]
SvenDowideitthere's only 1277 files in working/cache [12:48]
MichaelDaumhow many entries in the foswiki.xxx tables? [12:49]
SvenDowideitone mo (or 4 [12:49]
gac410AlexisHazell: You can update f.o/System, but odds are the changes will just disappear on the next install. It's a lot of work to 3-way diff every modified System topic with previous distribution, and new distribution, to figure out changes. [12:50]
SvenDowideitdeps has 88,971
pages has 7,248
so its not been runnig that long :)
[12:50]
gac410And then decide if it needs to go into svn, update, and rebuild the release. [12:50]
AlexisHazellgac410: *nod* i've just changed core/data/System instead. [12:50]
gac410Yes please. They will get put onto f.o hopefully today ... though at last report Lavr was only half way through updating tasks. [12:51]
MichaelDaumSvenDowideit, hm. that needs debugging. [12:52]
MartinCleaverHey Babar - any further thoughts on your Redis idea http://irclogs.foswiki.org/bin/irclogger_log/foswiki?date=2011-05-05,Thu&sel=317#l313 [12:52]
SvenDowideitMichaelDaum, sure does - i don' thave _that_ many topics on here yet
anyhoo - time for bed - nite all
[12:53]
BabarMartin: If you wouldn't have pasted the logs, I wouldn't have remembered I even mentionned it ;) [12:53]
MichaelDauma ?refresh=all would nuke the page cache as well...to start gathering stuff from scratch
nite
[12:53]
foswiki_irc4hi everyone, was wondering if someone could a newbie a hand in setting up mail notifcation, i think i need to setup a cron job but have no idea how to do so and help isnt as hand holding as i want it to be [12:54]
AlexisHazellfoswiki_irc4: Happy to try to help, if i can. :-)
foswiki_irc4: What aspect of setting up the cron job do you need help with?
[12:56]
MartinCleavery, Babar it was the only reference, so figured as much and pasted the log entry ;)
I have a client saving lots of info using (TWiki's) SetGetPlugin - but I note that the plugin not concurrent safe.
I've been using redis with node.js and figured it might be appropriate
[12:57]
foswiki_irc4hey alexis, i have no clue what a cron job is i just want to setup a simple weekly notifcation for all webs [12:58]
AlexisHazellfoswiki_irc4: What OS is your web server running on? [12:58]
foswiki_irc4im using windows server 2003 [12:59]
AlexisHazellfoswiki_irc4: Ah, i'm afraid i'm not familiar with setting up notifications on a Win system .... "cron jobs" are something on Unix-style systems where a particular command gets run at user-defined intervals.
foswiki_irc4: Basically you need to find a way of telling Win to run the tools/mailnotify script once a week, for the web(s) you want notification changes for.
[13:00]
gac410foswiki_irc4: On windows it's the Task Manager in the Admin panel. (I don't know much beynod that) [13:01]
foswiki_irc4ah am i supposed to tell it to run a scheduled task? [13:04]
gac410yes. That's what cron is on linux [13:04]
foswiki_irc4cool ill have a look and see what i can do [13:04]
Babarhey guys, that's documented :p
wish I could find the docco back
[13:09]
gac410AlexisHazell: Doc changes like Item11718 ... you can just check in changes in the "catch-all" doc tasks shown at the top of every task via the view template. Item11383 for 1.1.5
save yourself a bit of work opening tasks. :)
[13:10]
AlexisHazellgac410: Thanks, i wondered about that, decided to err on the side of caution. :-)
(i.e. wasn't sure if some doc changes were more suitable for 1.1.5 than others.)
[13:11]
Babarhttp://foswiki.org/System/SiteTools#Setting_the_automatic_e_45mail_schedule ok, I'm sure I saw a better doc, but can't find it :( [13:12]
gac410Unless it's a proposal to completely restructure the docs or some other major rewrite, small incremental changes can go in via the general doc tasks.
Babar, maybe the support docs on installing on windows?
[13:13]
Babarok, can't find it :( [13:15]
foswiki_irc4hmm, only issue ive got is mailnotfy is not a program for it run , so wondering do i have to get it to run in perl? [13:21]
gac410foswiki_irc4: yes, on windows, you'll need to use perl -I ..\lib mailnotify or something like that. [13:23]
foswiki_irc4okay ill have a go [13:23]
...... (idle for 25mn)
***ChanServ sets mode: +o OliverKrueger [13:48]
gac410Pling Lavr, are you around? [13:48]
***OliverKrueger sets mode: +oooo Colas gac410 Lynnwood pharvey
OliverKrueger sets mode: +oo ThomasPHaeck AndreU
[13:54]
.... (idle for 16mn)
OliverKruegerMichaelDaum: I got a WikiWorkbench beginner question. :)
I want to create a new TopicType which has some domain specific form fields and inherits everything from ClassifiedTopic.
I create a new WikiApp and in there a new TopicType. Can I express the inheritance between my new type and the classifiedtopic across webs/apps?
[14:10]
MichaelDaumyes [14:10]
foswiki_irc4just wondering do you have to rename the mailnotify file? at the moment it doesnt havent a file extension [14:11]
OliverKruegerMichaelDaum: Do I have to combine the formfields manually? (Category, Tags plus my own fields) [14:11]
MichaelDaumyes. there's no magic inheritance in DataForms unfortunately. so we must use manual mixins. [14:12]
OliverKruegerok, thx. :)
foswiki_irc4: tools/mailnotify does not have an extension like ".pl".
foswiki_irc4: but you can rename it to whatever you like, if your OS needs that.
[14:13]
foswiki_irc4thanks oliver, just needs some handholding trying to setup web notifcations on windows server 2003, i kno i have to do it through scheduled tasks [14:16]
OliverKruegerOliverKrueger runs away. [14:16]
foswiki_irc4lol [14:16]
Babaryes, you're on the right track
unfortunately, most of us haven't used windows in a very long time (thank God for that)
[14:16]
OliverKruegerfoswiki_irc4: accept my deep sympathy. ;) [14:17]
Babarwhich remidns me... I should boot my VM to send an email... [14:17]
***OliverKrueger sets mode: +o MartinCleaver [14:18]
foswiki_irc4hmm whats command line i have to run in perl to get it to run the mailnotify [14:23]
gac410change to the tools directory, and then perl -I ..\lib mailnotify [14:25]
OliverKruegerfoswiki_irc4: are you talking about the shebang?
(the first line of the mailnotify script)
[14:25]
gac410Windows doesn't use the shebang does it? (Except under apache) [14:25]
foswiki_irc4i do have apache installed [14:26]
OliverKruegergac410: not sure, but I've seen !#c:/foo/bar lines. [14:26]
foswiki_irc4ive used the fsowiki windows installer [14:26]
gac410What I meant was apache uses the shebang to run the bin scripts, but cron / windows shell doens't use it for things like mailnotify [14:26]
OliverKruegergac410: k [14:27]
gac410foswiki_irc4: I just tested it on an old XP system I have. Change to tools directory. And then issue "perl -I ..\lib -I ..\bin mailnotify" the -I stands for Include - tells perl where to find library files. For some reason I needed to have both bin and lib specified here.
Hm. works fine without the -I ..\lib So just need perl -I ..\bin mailnotify
[14:37]
...... (idle for 29mn)
Lavrgac410 I am around
Just came back and continue walking through fixed bugs. Nothing dramatic found. Changed one bug to enhancement only.
[15:07]
gac410Okay great. Thanks Lavr, Have you had a chance to test? [15:14]
LavrI have tested around a little bit also. No issues seen yet [15:15]
gac410What is your process btw ... Do you normally re-generate the release notes after updating tasks? I had been just fixing up the Release Notes themselves
Oh, and for something like Item11312, where I've checked in a workaround, but not a real fix, would you clone the task, or just leave it open for 1.2
[15:16]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [15:16]
LavrI normally generate the bug list in the rel note just before release. Otherwise It is an endless chase
I often create a new task for the remaining thing to fix and update the title of the old to describe just what was checked in
[15:16]
gac410okay - I'll do that then. The workaround to exclude WYSIWYG when native html tables are in the topic is not really a fix. [15:18]
LavrAre you saying we disable the Wysiwyg editor when we have an HTML table now? [15:19]
gac410I wasn't sure which was more work, regenerating the release notes, or updating the version I built for the RC since there are so few changes. Actually ReleaseNotes isn't as big a deal as regenerating the bullet list for the Download task ... which doesn't have a helper
Please post / email / whatever when you think you are done and I can release I have to run out for a bit in a little while Dentist :(
[15:20]
LavrI need to understand that Item11312. If it is what I understand then it breaks a lot of the common use we have.
We paste in tables from MS programs often and they end up as HTML table. If that then disables Wysiwyg editing you are fucked.
[15:22]
gac410It can be overridden by setting a WYSIWYG_EXCLUDE in site preferences. The issue is that embedded white space inside HTML Table tags breaks things.
pharvey tried to fix it but was not successful.
[15:25]
LavrIt is a horrible workaround of a pasted in table which is saved causes the editor to be disabled leaving an awful raw html table as the only means to edit. I would not upgrade to 1.1.5 with such a bug [15:26]
gac410I think the only other solution for 1.1.5 will be to revert WYSIWYG, TMCE, or delay 1.1.5. [15:27]
LavrOne or the other. It is not a good work around to disable the Wys editor when you have html tables.
People will be trapped in that daily.
And they probably have html tables from MS programs without any empty lines in them so they would never see the original problem.
[15:27]
gac410yes, It's pretty ugly. [15:29]
LavrWait with the release then. People are better off with a few more days of 1.1.4 than living with this [15:30]
gac410The good news is that SvenDowideit and pharvey did get a test case written that exposes the issue in the TML2HTML code in WYSIWYG plugin [15:30]
LavrThat is good. Then it may be a matter of 1-2 days till we have a fix.
I am on vacation now ALL of April.
[15:31]
gac410I am not too confident on that. The WYSIWYG conversion of TML to HTML ... esp. the encoding of the whitespace ... is really a bit of a rats nest.
I think this all starts with mtempest changes to better encode whitespace, so that some more advanced features of tmce can be used. It works for lists, but has issues in tables.
[15:32]
LavrYes. I remember the original bug where Rich suggested preserving newlines when doing Wysiwyg and I warned against it.
I am now at the enhancements. And I turn many of them into bugs. What a developer sees as a code enhancement is invisible to the users or in some case seen as bugs from the users.
[15:37]
.... (idle for 15mn)
I just checked several topics on the local copy I have of work Foswiki. We have MANY topics that suddenly cannot be edited and reveal horrible HTML tables to the editing user.
We cannot release 1.1.5 with that workaround. It is impossible.
I also found an old (tm)wiki bug 5076 where we were in similar situation where TML in HTML tables goofed things up. We just have to be able to handle that. And if we broke it we need to fix it or revert the code that broke it.
I am done with the bug walk through.
[15:55]
foswiki_irc1Hello all - I am trying to upgrade from 1.1.2 to 1.1.4 and I'm having issues. I am following the instructions as per here: http://foswiki.org/Download/FoswikiRelease01x01x04#Installation [15:58]
LavrDid you use the upgrade pack? [15:59]
foswiki_irc1I am using Foswiki-upgrade-1.1.4.tgz [16:00]
LavrWHat problems? [16:00]
foswiki_irc1Well I copy it to /var/www/Foswiki-1.1.2 and then unzip over as stated in article
I restart apache and I receive errors. Hold a moment and I'll post exact error.
first I get a warning that port 80 has no virtual hosts when I restart service
Then when hitting on port 80 I get ERROR: Could not open logfile /var/www/foswiki/working/logs/error.log for write. Your admin should 'configure' now and fix the errors! at /usr/local/share/perl/5.10.1/CGI/Carp.pm line 354
[16:00]
LavrIs your Foswiki tree owned by Apache user? (httpd, apache, www-user depending on distro)
When you copy over the tgz you may end up with files and directories owned by root.
E.g. when I add files as root I have to do a "chown -R apache:apache /var/www/foswiki" on my RedHat.
Redhat runs the httpd daemon under the username "apache" as you can guess.
and group apache
[16:03]
fsfsFoswikiBot: seen ArthurClement [16:07]
FoswikiBotfsfs: Sorry, I haven't seen ArthurClement. [16:08]
fsfshmm, what irc handle do I need to ask for?
FoswikiBot: seen ArthurClemens
[16:08]
FoswikiBotfsfs: ArthurClemens was last seen in #foswiki 3 days 4 hours ago joining the channel. [16:09]
fsfsmuch better :-) [16:09]
....... (idle for 30mn)
***ChanServ sets mode: +o OliverKrueger [16:39]
foswiki_irc5I got dropped before
Sorry I didn't respond I just checked log and saw Lavr responded to me. Anyway I used the upgrade script and managed to get it working properly.
using the foswiki-upgrade-check that is.
[16:44]
.......... (idle for 47mn)
gac410Ping Lavr, you might also want to review Item11603 / Foswikirev:14208 This fix also disables WYSIWYG [17:32]
FoswikiBothttp://foswiki.org/Tasks/Item11603 [ Item11603: Protect inline script and style tags from wysiwyg ] http://trac.foswiki.org/changeset/14208 [ Changeset 14208 – Foswiki ] [17:33]
.... (idle for 16mn)
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/z7_FZw
[foswiki/Release01x01] Item11312: Revert making table a default EXCLUDE - GeorgeClark
[17:49]
***GithubBot has left [17:49]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [17:49]
GithubBot[foswiki] foswiki pushed 1 new commit to master: http://git.io/8s2RAw
[foswiki/master] Item11702: $status regex was too rigid, Reason-Phrase (*<TEXT, excluding CR, LF>) may be included - FlorianSchlichting
[17:50]
***GithubBot has left [17:50]
FoswikiBothttp://foswiki.org/Tasks/Item11702 [ Item11702: Redirect call in Save and Rename have incorrect call to Foswiki::redirect() ] [17:50]
...... (idle for 29mn)
gac410The problem with the TML2HTML routine is that <table triggers the push of a <p> ahead of the table and sets "InParagraph", then the first blank line inside the table triggers </p> which breaks the table. However my attempts at a fix breaks it worse :( [18:19]
...... (idle for 27mn)
Okay ... have a fix for the broken table issue. Somewhat brute force, and it breaks one other unit test, which expects tables to be embedded inside a paragraph <p> ... the table ... </p>
Which I don't think makes sense.
[18:46]
I can change each blank line to either <p></p> or <br/><br/> when inside a table. WYSIWYG will save it as a TML table with the embedded <> markup. any opinions out there? [18:57]
..... (idle for 22mn)
GithubBot[foswiki] foswiki pushed 1 new commit to Release01x01: http://git.io/X3IRZA
[foswiki/Release01x01] Item11312: Minimal processing of HTML tables - GeorgeClark
[19:19]
***GithubBot has left [19:19]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [19:19]
gac410This fix needs careful review! [19:22]
..... (idle for 21mn)
foswiki_irc5Should www-data or root have ownership over foswiki directory? [19:43]
gac410It depends upon your distribution. often www, or www-data, or apache. generally never root. [19:44]
foswiki_irc5hmm
I'm getting permissions errors that the pub directory is not writable
[19:44]
gac410what distribution / OS? [19:45]
foswiki_irc5I set permissions as recommended to 755
ubuntu / 1.1.4
[19:45]
gac410I think on ubuntu, the ownership is "www", and group is "www-data" Directories 755, files 644.
Unless you are configured apache for SuexecUserGroup to override the userid of the CGI scripts.
[19:45]
foswiki_irc5oh no. I'm keeping it really basic [19:46]
gac410Hang on . I don't use ubuntu myself, but I think there is a support topic. [19:46]
foswiki_irc5I did find one here: http://foswiki.org/Support.SettingFileAccessRightsLinuxUnix
but the foswiki version is 1.0 not 1.1
I've tried everything in that article to no avail
[19:47]
gac410No differences between 1.0 and 1.1 for file ownership that I know of
debian/Ubuntu/Kubuntu : chown -R www-data:www-data /path/to/foswiki
So owner and group should be www-data
[19:47]
foswiki_irc3got booted.
let me look at chatroom log
[19:49]
gac410I'll replay
(03:48:04 PM) gac410: debian/Ubuntu/Kubuntu : chown -R www-data:www-data /path/to/foswiki
(03:48:12 PM) gac410: So owner and group should be www-data
[19:49]
foswiki_irc3yes that is what I have.
thanks
I used the foswiki-upgrade-check to upgrade
[19:50]
gac410Where do you see the errors? In configure, or when you try to attach a file to a topic? [19:50]
foswiki_irc3in configure
Error: 354 directories or files have insufficient permissions. Insufficient permissions could prevent Foswiki or the web server from accessing or updating the files. Verify that the Store expert settings of {RCS}{filePermission} (0644) and {RCS}{dirPermission} (0755) are set correctly for your environment and correct the file permissions listed below.
First 10 detected errors of insufficient, or excessive permissions
[19:50]
gac410Just paset one error please
Or use pastebin.
[19:51]
foswiki_irc3ok
It's complaining about the /var/www/foswiki/data directory. That it is not writable
[19:53]
gac410hm. what does "ls -lad /var/www/foswiki/data" show for permissions and ownership? [19:54]
foswiki_irc3drwxr-xr-x 9 www-data www-data 4096 2010-11-28 23:43 /var/www/foswiki/data [19:54]
gac410well that sure looks okay. Have you tried to attach a file to a topic to see if that part of foswiki is functional? [19:55]
foswiki_irc3I haven't [19:56]
gac410And could you pastebin (http://pastebin.com) the complete block of error messages from configure. That might give me some hints.
Configure can get confused down in the permissions checking - though not often. I thought 1.1.4 was pretty solid.
[19:56]
foswiki_irc3http://pastebin.com/VSGR33uM [20:00]
gac410Ah... it's the files themselves that are not writable. Could you do a "ls -la /var/www/foswiki/pub/System/JavascriptFiles/foswikiStyles.js.gz"
(or any of the files there? Just one.)
[20:01]
foswiki_irc3k
-rw-r--r-- 1 root root 20 2012-04-02 12:53 /var/www/foswiki/pub/System/JavascriptFiles/foswikiStyles.js.gz
[20:01]
gac410yeah. The upgrade changed the ownership of the changed files to root. It's usually safe to just do a chown -R www-data:www-data pub
and same for data
I use that script to upgrade foswiki.org. My usual procedure is to run the upgrade, then chown -R... the entire foswiki directory, pub, data, lib, etc.
[20:02]
foswiki_irc3I could swear I did run chown -R on whole dir [20:04]
gac410as root?
Have to either sudo or su to change the ownership away from root and back to www-data
[20:05]
foswiki_irc3o
that did the trick!
[20:05]
gac410Great! [20:05]
foswiki_irc3thanks so much! [20:05]
gac410yw [20:05]
............ (idle for 59mn)
GithubBot[foswiki] foswiki pushed 2 new commits to master: http://git.io/jySGCA
[foswiki/master] Item11312: Revert making table a default EXCLUDE - GeorgeClark
[foswiki/master] Item11312: Minimal processing of HTML tables - GeorgeClark
[21:04]
***GithubBot has left [21:04]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [ Item11312: TinyMCE Corrupting HTML Tables ] [21:04]
............... (idle for 1h12mn)
***ChanServ sets mode: +o OliverKrueger
OliverKrueger sets mode: +oooo terceiro uebera|| Rich_Morin MartinCleaver
OliverKrueger sets mode: +o Colas
[22:16]
............. (idle for 1h1mn)
gac410howdy pharvey - I think I've got a working fix for Item11312 ... rather brute-force-ish. but it doesn't corrupt tables at least with the simple ones in the test cases. [23:17]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [23:18]
.... (idle for 16mn)
dj_segfaultHi, folks. Is there a way to set attributes on an individual cell in a table (like background color or border)? [23:34]
pharveyhi gac410, I did see that - your approach is a perfectly valid one, it just neuters the whitespace preservation within tables (I had an idea I mentioned in the task, but it's more risky, your way is simpler). [23:35]
Babaryes, with JQuery
dj_segfault: Micha has some examples... JQgrid? don't remember which JQ plugin does that easily
[23:35]
pharveyBabar, really? Did he doc that somewhere? [23:36]
dj_segfaultHm... OK I'll search for that [23:36]
pharveydj_segfault: it's a very common request, I wish somebody would create a brainstorming topic with some ideas for new syntax. My first thought was |.red content |.blue content|, but meh :) [23:36]
gac410pharvey - My assumption (maybe invalid) is that other than blank lines, whitespace inside pure html markup is ignored, so we should not have to preserve it. Maybe not, but I think that's fair. If I'm writing real html I'd expect to have to use &nbsp; etc. [23:37]
pharveyjqgrid doesn't work with TML tables, AFAIK
gac410: I'm thinking more of the newline removal, after somebody goes to the effort of indenting/newline-ing their tables.
[23:37]
dj_segfaultThat would be pretty inconsistent with how other things are done in tables. Maybe %CELLATTR{" ..."}% [23:38]
pharveyand where would you put the syntax?
in the cell?
ok
[23:38]
dj_segfaultIt would have to be [23:38]
pharveyusually when I have this conversation with my own users or on IRC, they're just wanting background colour [23:39]
gac410Ah. yeah, though that would have to prevent conversion from html to tml. 1.0.10 behavior and I think 1.1.4, converts to tml if the table doesnt' use attributes tml can't support. [23:39]
dj_segfaultI haven't yet studied how the tables get turned into HTML (laws and sausages and HTML), so I can't say where the easiest to implement it would be, but anything other than putting it in the cell will require row/column addressing which gets confused by sorting. [23:40]
gac410gac410 took one foray into TablePlugin ... and hopes to NEVER return.
pharvey, my fix is *mostly* similar to 1.1.0 and 1.1.4 which will convert the html table to | A | B <p> </p> <p>C</p> | D | (Note that C is a separate paragraph. My 1.1.5 fix results in | A | B <p> </p> C | D | (omitting the paragraph markup around the C
[23:41]
pharveyok [23:45]
dj_segfaultWell, almost everything I'm using Foswiki for involves crazy table stuff. [23:46]
gac410dj_segfault: There are 2 implementations of Table rendering (ignoring edit!) default render can output a simple html table, no sorting, etc. TablePlugin overrides the rendering and adds sorting, etc.
WysiwygPlugin has it's own version :P
[23:47]
pharveyJQTablePlugin does what TablePlugin does, but in javascript. Which is kinda nice
I started work on Foswiki::DOM to unify this mess. It's not... going quickly :)
[23:47]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::DOM [23:48]
pharveydj_segfault: FWIW JQGridPlugin is really nifty, but requires a topic-per-row paradigm. Eg. http://wiki.trin.org.au/Mangroves/Bibliography/ [23:48]
dj_segfaultLooking for JQTablePlugin now [23:48]
pharveyThen there's EditRowPlugin, which is a really nice ajax-tastic version of EditTablePlugin. I think there's even a proposal to replace EditTable for Foswiki 1.2 [23:49]
dj_segfaultpharvey: I don't see any tables on that page so I don't know what you mean by topic per row. [23:50]
pharveyhttp://foswiki.org/Development/ReplaceEditTablePluginWithEditRowPlugin [23:50]
gac410I don't like Sven's fix for script and style tags. Disabling wysiwyg. the "Palatable Tags" code protects the start and end tags, but renders the stuff between them It *really* ought to protect the entire block. But like with Tables, disabling wysiwyg is a bit draconian
I don't think we want to loose the conversion of <table> to | tml | ... Preserving indenting of the <tr> <td> and all that would mean preserving the HTML. And there are some huge wiki application advantages to allow tables to convert to tml.
[23:50]
pharveydj_segfault: did it come up eventually? It's supposed to look like http://imagebin.org/206488 [23:52]
dj_segfaultNo. I just saw paragraphs [23:53]
pharveyIt's supposed to be like an ajax SEARCH
Interesting.
which browser?
[23:53]
LavrI use the conversion from HTML to TML table often.
Hello guys
[23:53]
pharveyhi Lavr [23:53]
gac410Howdy Lavr [23:53]
dj_segfaultAh, NoScript stopped it from working [23:53]
LavrWhen you copy a table from word or excel it is always HTML with lots of garbage
And many just use that abd
And edit it as Wysiwyg
[23:54]
gac410Lavr, I think I have Item11312 fixed. checked in to trunk and release11 branch. So you could try a copy/paste on trunk. [23:54]
FoswikiBothttp://foswiki.org/Tasks/Item11312 [23:54]
LavrBut I always remove all content of the table html tag and the header tags
And then let TMCE save it
And then it turns into clean TML
If I was smarter I would make that a button feature in the editor :-)
[23:54]
gac410Sven checked in a fix that does the same disable wysiwyg if <script> or <style> tags are detected [23:56]
pharveyI've thought about ways of doing the cleaning automatically. TinyMCECleanStickyBitsPlugin tried to. It added a new toolbar button to clean those attributes from selected text.
I wonder if TinyMCECleanStickBIts still works
[23:56]
LavrI will try the fix tomorrow. I am in bed now with my ipad and sleeping wife
I am not concerned about script. People that add scripts to a page are geeks anyway. But the table is something normal users paste in all the time
[23:56]
gac410okay. I'll be tied up tomorrow and Wednesday, don't know if I'll get to build 1.1.5 then until text week [23:58]
LavrStyle tags may be a problem
It is ready when it is ready. It is a great job done so far with all these bug fixes
[23:58]
gac410And though it's not supposed to be a feature release, there is a bit of creep in there as well. [23:59]

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