#foswiki 2014-04-16,Wed

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

WhoWhatWhen
***ChanServ sets mode: +o pharvey [00:41]
...................................... (idle for 3h8mn)
ChanServ sets mode: +o SvenDowideit [03:49]
............. (idle for 1h0mn)
gac410 has left [04:49]
..... (idle for 21mn)
harlanHow can I get a '|' character to appear as text in a table cell? [05:10]
............. (idle for 1h2mn)
***ChanServ sets mode: +o CDot [06:12]
................... (idle for 1h34mn)
GithubBot[foswiki] FoswikiBot pushed 2 new commits to master: http://git.io/ahLJEQ
foswiki/master 7c4d0e0 CrawfordCurrie: Item9693: apostrophical error in error message...
foswiki/master b728135 CrawfordCurrie: Item9693: apostrophical error in error message...
[07:46]
***GithubBot has left [07:46]
FoswikiBothttp://foswiki.org/Tasks/Item9693 [ Item9693: Documentation updates for Foswiki 1.2.0 ] [07:46]
................................. (idle for 2h44mn)
***ChanServ sets mode: +o Lynnwood [10:30]
MichaelDaumharlan, use %VBAR% ... defined in System.DefaultPreferences [10:31]
.................. (idle for 1h29mn)
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/gnomlA
foswiki/master 7efe7e5 MichaelDaum: Item12859: brought back thumnail image type...
[12:00]
***GithubBot has left [12:00]
FoswikiBothttp://foswiki.org/Tasks/Item12859 [ Item12859: ImagePlugin: type="thumb" breaks rendering of image ] [12:00]
rickb_Is anyone on who has experience with SolrPlugin? I'm having trouble getting Foswiki (1.1.9) to display results. The Solr server is definitely working. I can query using curl and get results, but nothing in Foswiki. [12:00]
MichaelDaumhi rickb_ [12:00]
rickb_Hi MichaelDaum [12:00]
MichaelDaumthere are a couple of things to check [12:00]
rickb_lay 'em on me! [12:01]
MichaelDaumfirst, let's see if you can search from the cmdline, how about that? [12:01]
rickb_yes. using curl. I get lots of xml output [12:01]
MichaelDaumno, plain command lining. cd <foswiki-dir>/bin; ./rest /SolrPlugin/search q=test [12:01]
rickb_ah!
I will try that. I cannot at the moment, unfortunately.
I need to start a vpn which will cause this session to drop.
[12:02]
MichaelDaumbefore you do that
have you been able to run the indexing process?
[12:03]
rickb_I appreciate your help. I will have to get back to you with any results. In fact, I have to go very soon. If you have a couple of other suggestions, please give them quickly [12:03]
MichaelDaumcd .../tools; ./solrindex topic=System.WebHome [12:03]
rickb_yes. I ran indexing and have document from the wiki webs in the index [12:03]
MichaelDaumnext ... when you are back here. try opening the firebug console (or chrome equivalent) ... watch out for javascript errors
there might be an error in your rest setup
[12:04]
rickb_ok. I have to go. I'll check and get back. thanks! [12:09]
MichaelDaumyw [12:09]
........ (idle for 37mn)
***ChanServ sets mode: +o gac410 [12:46]
..... (idle for 24mn)
rickb_Hi.
I see MicahelDaum is no longer on.
I'm new to IRC. Not sure if it's polite to follow up when the person I was talking to is no longer on.
Anyone else familiar with SolrPlugin?
Or with JQueryPlugin?
It looks like that may be the culprit.
[13:10]
gac410Hm... MichaelDaum is here ... just doesn't have oper status right now. Yes you can follow up even if someone is not on. [13:13]
***gac410 sets mode: +o MichaelDaum [13:13]
MichaelDaumrickb_, no I am here [13:13]
rickb_When I opened Firebug, it shows a JSON.parse error
Oh. Hi!
So, search from CLI worked.
[13:13]
MichaelDaumokay now have you logged in or are you visiting the site as a guest? [13:14]
rickb_Firebug shows the JSON.parse error [13:14]
gac410rickb_: The channel is logged. Many will check the logs to look for follow-up so you can always comment even if someone is not online. [13:14]
foswiki_irc8@gac410, i left for the day around the same time you left for dinner. before i left, i checked the email settings in my configure tool and the drop down for mail method was blank, and i wasnt able to change the setting. i saved toggled some other minor setting off and on, saved changes, went back to the page, and was able to change it to SMTP
verified via grepping the cfg file per your instructions. seems to be working....thought I havent changed the .pm file back - do you think it's worth a shot?
[13:14]
MichaelDaumrickb_, it seems your server doesn't return proper JSON. [13:15]
rickb_gac410: thanks. [13:15]
gac410foswiki_irc8: sure. It *shouldn't* fail if MailMethod is set. But then again it won't do much either way. It just goes some consistency checking. [13:16]
MichaelDaumthis could be the case due to rest still being listed in {AuthScripts} in your LocalSite.cfg [13:16]
foswiki_irc8gac410: okay will do. Any ideas why changing my password method from HtPasswd to LdapPasswd might have made my site unusable? is that likely to be an ldap configuration error on my part? [13:17]
MichaelDaumfoswiki_irc8, enable $Foswiki::cfg{Register}{AllowLoginName} = 1; [13:17]
gac410foswiki_irc8: I don't use ldap and don't know for sure. When you say "unusable" did it crash, or just not allow login. Yay.. MichaelDaum answered. He knows a LOT more about ldap than I do :) [13:18]
MichaelDaumthere is an error in the core and/or the LdapContrib where Foswiki goes into a deep recursion otherwise
the other cause of the slowdown is that your ldap directory simply isn't responding, or responding too slow.
so the initial warmup for the ldap cache in Foswiki isn't carried out.
[13:18]
foswiki_irc8interesting. is there a nice way to test that the ldap connection is working and valid, prior to committing changes via the config tool? [13:20]
MichaelDaumtry to do so on the cmdline once you've set {AllowLoginName} = 1 ... using cd <foswiki-dir>/bin; ./view refreshldap=on >/dev/null
foswiki_irc8, yes. use the ldaptest script in <foswiki-dir>/tools
this is a small perl script that you need to edit and customize at the top of it
[13:20]
gac410MichaelDaum: Did you see the question about TopicInteractionPlugin and WidgetsSkin? There is no explicit dependency on NatSkin in that plugin, but it does have lots of other dependencies. [13:31]
MichaelDaumyep.
it does fine with PatternSkin...so
[13:31]
gac410I just gave a lame answer - basically an "i don't know" Ah... that's a good sign.
I played around with WidgetsSkin a few years ago, but never had any success :)
[13:32]
MichaelDaumtoo bad
the theme site it points to seems to be gone as well
[13:33]
rickb_MichaelDaum: When you say the server doesn't return proper json, do you mean solr server? [13:34]
gac410Ah. Unfortunately we need a project wide "scrub" of the extensions. We have way too many that are languishing away - not working with recent perl (qw issues) etc. It looks bad for the project overall. [13:34]
MichaelDaumno the foswiki server which runs as as proxy to the solr underneath [13:34]
rickb_OK. Could that be the JsonPlugin? Maybe a version problem? It's pointing to https://<servername>/foswiki/pub/System/JQueryPlugin/jquery-1.8.3.js
Sorry, JQueryPlugin
[13:36]
MichaelDaumrickb_, hm, solr doesn't use any JsonPlugin
it does a http://.../bin/rest/SolrPlugin/proxy??fl=... reading the net results from solr ...
[13:43]
rickb_MichaelDaum: Interesting. That's what's being implicated in FIrebug. [13:44]
MichaelDaumtry to grab the /proxy?... url from firebug and open it in a new tab of your browser [13:45]
rickb_My server is set to require https - could that be the problem? It also uses a self-signed cert. [13:45]
MichaelDaumnope. mine is https too (using nginx and this spdy :))
rickb_, did you remove rest from {AuthScripts} already?
[13:45]
rickb_MichaelDaum: I have not removed rest from {AuthScripts}
MichaelDaum: When I go to this URL: https://<servername>/foswiki/bin/rest/SolrPlugin/proxy?? I get no errors and lots of json
[13:47]
MichaelDaumgood. could still be it is a json encoded error message.
but first remove rest from AuthScripts
[13:49]
rickb_Nothing looks like an error. I recognize document contents from my Webs
At the top, it looks like this: {"responseHeader":{"status":0,"QTime":8},"response":{"numFound":5114,"start":0,"docs":
I can remove rest. Just curious why?
OK I removed rest from {AuthScripts}.
MichaelDaum: Went back to this URL https://<servername>/foswiki/bin/view/System/SolrSearch and get the same JSON.parse error: SyntaxError: JSON.parse: expected ',' or '}' after property value in object
[13:50]
MichaelDaumrest should be removed from {AuthScripts} so that we won't get a http redirect to a login screen instead of json [14:02]
.............. (idle for 1h5mn)
rickb_At this point, I need to figure out how to debug this. Any ideas? [15:07]
gac410rickb_: You might cut/paste the json into something like http://www.freeformatter.com/json-validator.html to see where the json has issues. [15:10]
rickb_gac410: thanks. I did that with the json returned by https://<servername>/foswiki/bin/rest/SolrPlugin/proxy??
gac410: I got "The JSON input is valid in JavaScript."
gac410: and "The JSON input is valid according to RFC 4627 (JSON specfication)."
[15:11]
gac410hm... Have you tried a different browser .. maybe it's a browser issue?
gac410 doesn't do much with json or javascript, so he's just guessing here...
[15:13]
rickb_Firefox and Chrome both show the same blank frame, with header, sidebar and footer, at https://<servername>/foswiki/bin/view/System/SolrSearch [15:15]
gac410rickb_: Well that exhausts my ideas. I've never installed solr, so I have no idea what's going on. [15:16]
rickb_Chrome gives a slightly different error in its javascript console: Uncaught SyntaxError: Unexpected token B
thanks for trying!
[15:16]
CDotgac410: I have a couple of questions about "new" configure. Any chance you might be able to answer? [15:28]
gac410I can try :) [15:28]
CDotok. question 1. The existing checkers have a function called "check". The changes added a function called "provideFeedbck". Why?
they appear to cross-call to eachother, in a spaghetti/tangled string sort of a way
[15:29]
gac410Ah... I think that's for the rest style checks. [15:29]
CDotwhat's the difference? I see fome check functions reference the query - an absolute no-no by the way - was that the reason? [15:30]
gac410no idea. Timothe did all that work. I did a lot of testing for him though. [15:31]
CDotthe checkers were designed/intended to be used for checking the content of Foswiki::cfg - no more, no less. They should never do anything "active". [15:31]
gac410The objective of all the work was to check before committing to the configuration and breaking the site. That is extremely valuable. I've killed several sites when saving which then caused an error that checker would have found. [15:32]
CDotok, question 2. Why was everything put into the Foswiki package, and then import used to spill the globals all over the code? [15:32]
gac410No idea. [15:32]
CDotok. yes, the check should happen before the config is saved, no question about that
it's not the *timing* of the check I have issue with, it's the splitting up into "check" and "provideFeedback" - I just don't understand the "why".
[15:33]
gac410I'll have to go look. I'm pretty sure there was a reason. :) [15:33]
CDotok. The code I'm working on has changed a *lot* from trunk, so I may not be seeing the wood for the trees
sadly Timothe is not a believer in documentation
or maybe that was "next".
[15:34]
gac410I doubt it. I'm afraid that's a trait of a lot of us. The code seems "obvious" at the time doc later gets lost. :( [15:35]
CDotsure, yes. I'm well aware that I'm the only person who writes obvious code. :-) [15:36]
gac410Timothe tended to work mostly from email Let me see if I can find his descriptions of his checkins. He'd send me giant patches, I'd test, then he'd commit to SVN once it was all working.
Have you reviewed his feature proposals. He discussed a lot of stuff there as well.
[15:37]
CDotcripes
no, I haven't :-/ I hadn't realised there was stuff to read there
I sort of understand most of what he did - though the certs config I don't grok at all, it's almost as much code as the rest of configure
[15:38]
gac410Yeah The Certs stuff he was really fanatical about. He really was a huge proponent of certificate based authentication, including validating that the connected server was who it claimed to be. .. preventing DNS spoofing type attacks.
I pushed several times to hide the complexity, but he wanted it front & center, because everyone should use certs and validate all connections.
The email autoconfig stuff though was very good. Some platforms have long timeouts, but mostly email now is completely automatic. Finds the most secure connection, tells you if authentication is required, validates the password, etc.
When he was working on email, we had a couple of users who were pulling hair over "unable to send" issues. Really difficult to debug. So his auto-config was in reaction to some rather painful support issues.
CDot: see http://foswiki.org/Development/AJAXOnDemandCheckersForConfigure
[15:40]
CDotwell, n.m. I can work it out. What I'm a bit stumped by is the AUDIT addition to .spec. I can't find any doc, anywhere [15:44]
gac410Ah.... .AUDIT.. There is a new config tab. Moves some of the painful long-running checks into a separate area
Used for file permission checks, deep dependency checking (all core + all extensions) ..
That's all documented in Development/AJAXOnDemandCheckersForConfigure
[15:44]
CDotaha, that feature proposal has what I was llooking for!
thank you!
[15:45]
gac410yw.
He really did follow the process. Nobody objective, and he took it as FullSpeedAhead. Then when strong objections over the JS requirements cropped up, it was WW-III
er. Nobody objected.
[15:45]
CDotstill a bit confused as to the justification for it, though. ConfigurePlugin does pretty much all that is described there, but totally ignores the AUDIT param :-/ [15:47]
gac410Justification for Audit? [15:47]
CDoty [15:49]
gac410Ah... if you have not found it, in Foswiki/Configure/resource is a script make_gz that will rebuild all the gzips if you are changing the javascript ... Otherwise you need to clean and rerun pseudo-install to recompile. [15:49]
CDotyeah, I found that, thanks. I haven't touched the JS so far, I've been trying to repair the core perl - which basically involves decoupling the checkers again so they have no UI dependencies, and can be called from ConfigurePlugin. [15:51]
gac410Oh... and another tidbit. The js has version checking. It's important to bump the version when making changes, so that configure can tell you that it's running with a stale gz file, etc.
Otherwise there are mysterious "fail" things going on when the js gets out-of-sync.
All the Audit stuff would make good sense to move into a rest verb, and be accessible from an admin topic. Same for the log viewer stuff.
[15:52]
CDoty, I also found that
what were Sven's objections to the JS?
[15:53]
gac410he wanted configure to be usable from lynx or other telnet based "browsers" [15:54]
CDotwhich is basically the requirement I'm trying to realise. [15:54]
gac410I think Timothe's next major step would have been to move to sending single fields rather than posting the entire configuration. At least I think that was the direction he was headed.
But I think that involved making a copy of Foswiki::cfg to retain in the server session without saving it.
[15:58]
Unfortunately ... after Timothe left, we made one major change that hurt functionality but greatly improved performance.
Basically tabbing off of any field sent in a delayed post for a quick check. But it got horribly bogged down after multiple quick changes
We turned that off, but it broke the counter of pending changes
It was Foswikirev:16938 and Foswikirev:16939 where I disabled the on-the-fly checking
[16:12]
FoswikiBothttp://trac.foswiki.org/changeset/16938 [ Changeset 16938 – Foswiki ] http://trac.foswiki.org/changeset/16939 [ Changeset 16939 – Foswiki ] [16:17]
......... (idle for 41mn)
gac410I guess we need to reject the implemented proposal Foswiki:Development/RedesignLoggerAPI since it made Log::Dispatch a core dependency, which requires Class::Load [16:58]
FoswikiBothttp://foswiki.org/Development/RedesignLoggerAPI [ RedesignLoggerAPI ] [16:58]
gac410If we reject Class::Load it all goes away. :(
At least the modernized logger.
[16:58]
fsfs: I've removed LogDispatchContrib from the MANIFEST. You shouldn't need that for your unit tests any more. [17:05]
MichaelDaum_: Your changes to templates broke the HTML Validations. Your <meta tags are not closed.
Foswikirev:17427
[17:11]
FoswikiBothttp://trac.foswiki.org/changeset/17427 [ Changeset 17427 – Foswiki ] [17:12]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/fi_yFQ
foswiki/master 8cd72d0 GeorgeClark: Item12104: Remove LogDispatchContrib from MANIFEST...
[17:15]
***GithubBot has left [17:15]
FoswikiBothttp://foswiki.org/Tasks/Item12104 [ Item12104: Implement Development.RedesignLoggerAPI ] [17:15]
gac410MichaelDaum_: Also, all of the links you added for the stylesheet css are missing the type='text/css' attribute. [17:21]
MichaelDaumHi gac410
which change is that?
argh. long day.
[17:24]
gac410Hi MichaelDaum I'm almost done fixing. :) Foswikirev:17427 It's why the 144 failures on HTMLValidationTests [17:25]
MichaelDaumthanks [17:26]
gac410yay,... running without failures... whoops. Still a few failures: Undefined subroutine &Foswiki::Contrib::JsonRpcContrib::dispatch called at /var/www/foswiki/trunk/core/test/unit/HTMLValidationTests.pm line 186. [17:27]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Contrib::JsonRpcContrib [17:27]
gac410* F: HTMLValidationTests::verify_switchboard_function_jsonrpc_default + pattern, plain and print [17:27]
CDotgac410: my architectural vision is somewhat different [17:28]
MichaelDaumtoo bad pattern isn't html5 [17:28]
CDotrather than maintaining a complete partial state in the server, I favour small incremental changes with immediate feedback [17:28]
gac410CDot: How do you address that a change on one variable might trgger a validation fail on another very different area. [17:29]
CDotexample? [17:29]
MichaelDaumin html5 link-rel-stylesheet don't need an extra text/css type. in the same vein script defaults to text/javascript ... [17:30]
gac410Unfortunately things are so interrelated that changing A can cause B to flag an error. I need to think abou tit. [17:30]
CDotI have (tentatively) recoded the .spec parser so that it builds a dependency tree for explicit dependencies
the reason I ask for an example is for implicit dependencies
[17:30]
gac410CDot I'll know when when I see one :D [17:31]
CDotmy gut feeling is that if we have implicit dependencies, we are doing something wrong
the basic idea is that by doing incremetal changes, a screwup becomes visible immediately
[17:31]
gac410Agreed. I'm pretty sure that there were a couple that we stumbled across. [17:31]
CDotand a "full check" is always still available (e.g. after an offline LSC edit) [17:32]
gac410CDot I do agree with you. [17:32]
CDotI dislike intensly this "fix" button. Configure is not a wizard, it's too dumb for that [17:32]
gac410Which fix button. [17:33]
CDotTimothe specced a "fix" button for e.g. to repair perms [17:33]
gac410Ah... yeah, I don't think it's implemented yet, is it?. That would be much better done as a separate rest command I suspect.
Permissions in general is a bit of a minefield. Checking even is nonsense on systems using AFS, for ex.
[17:34]
CDotright. We briefly discussed a wizard some time ago. But i see that as a separate beast. [17:36]
gac410But on the other side, if checker says, your site is borked because foswiki can't write to a log file, or a web directory, having a 'fixit' button is better than saying ... go fix it yourself. But agree, doesn't need to be integral part of configure. [17:36]
xdaxdbgac410: did you ever find a more basic example of a search topic dataforms via an html form? [17:40]
gac410xdaxdb: no, But basically the form needs to get the topic, setting URL params SomeTopic?formfield=blah And the %SEARCH references %URLPARAM{formfield}% to apply the field. [17:41]
xdaxdbyea, but i am more interested in dataforms than url or topic attributes [17:42]
gac410Okay MichaelDaum_ ... I've checked in fixes to the templates. All of the HTMLValidation tests now pass except for the jsonrpc tests.
But... the way to pass a field between a html form and a search of the dataform, is via a URL param. That's the connection. I have no idea if there is another way, I don't do much form work.
[17:42]
xdaxdboh ok i see [17:44]
GithubBot[foswiki] FoswikiBot pushed 1 new commit to master: http://git.io/3K4gnQ
foswiki/master 133df5e GeorgeClark: Item12192: HTMLValidation Tests are failing...
[17:45]
***GithubBot has left [17:45]
FoswikiBothttp://foswiki.org/Tasks/Item12192 [ Item12192: PatternSkin fixes and enhancements in trunk ] [17:45]
............................ (idle for 2h18mn)
harlanThanks MichaelDaum_ ! [20:03]
......... (idle for 43mn)
rickb_Is there another way to test SolrSearch? All I have been doing is going to https://<servername>/foswiki/bin/view/System/SolrSearch - where I get a blank page. How else can I test if Foswiki can display formated search results? [20:46]
..... (idle for 23mn)
***rickb_ has left "Leaving" [21:09]
............ (idle for 55mn)
gregg4567 has left [22:04]

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