#foswiki 2016-02-10,Wed

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

WhoWhatWhen
ssinfodhello
I have a question about email configuration
[01:49]
gac410hello ssinfod
What's your question
[01:50]
ssinfodI want to use gmail (smtp) but I can't figure the correct settings [01:50]
gac410Foswiki 2.x or 1.1.x [01:51]
ssinfod2.x [01:51]
gac410What platform are you on? Linux? [01:51]
ssinfodyes, linux, ubuntu 14.04
what would be the correct "mail method" ?
[01:51]
gac410You should not have to touch that. [01:52]
ssinfodok, so Net::SMTP [01:52]
gac410No no... You want to use autoconfigure. Don't touch the advanced stuff [01:52]
ssinfodI will quickly list my settings.. [01:53]
gac410On "Basic Settings & Autoconfigure. You fill in 4 fields [01:53]
ssinfodyes. [01:53]
gac410webmaster email, [01:54]
ssinfodmy email..
ex: ssinfod@gmail.com
SMTP Host: smtp.gmail.com:465
[01:54]
gac410smpthost set to smtp.gmail.com
get rid of the 465
just smtp.gmail.com
[01:54]
ssinfodok
then SMTP Username I used my gmail email address and the password is my gmail password.
[01:54]
gac410username is your gmail account ssinfod@gmail.com
yes. And then press autoconfigure
[01:55]
ssinfodwhich one ? [01:55]
gac410Any of them. all the same [01:55]
ssinfodI got a window: "Validation"
then it is saying:
Testing STARTTLS with NO host verification on submission port (587) Net::SMTP>>> SSL peer verification: off
etc..
[01:56]
gac410okay... and down at the bottom, it should show you the settings, ... unless it failed [01:57]
ssinfoda big text.. [01:57]
gac410highlighted in blue. [01:57]
ssinfod"SMTP configuration using smtp.gmail.com failed. Falling back to mail program"
just before..
"This configuration appears to be acceptable, but testing is incomplete.
Authentication failed "
[01:57]
gac410Okay. So it didn't like your userid or password [01:57]
ssinfodcan the Webmaster email and SMTP user name be the same ? [01:58]
gac410I just tested it with my gmail account and it worked fine. If you have two-factor gmail auth enabled, it might not work though.
Yes webmaster and gmail account can be the same
[01:59]
ssinfodit is strange I received an email from google..
"Sign-in attempt prevented"
"Someone just tried to sign in to your Google Account ssinfod@gmail.com from an app that doesn't meet modern security standards."
Is this the "two factor" you are speaknig about ?
[02:01]
gac410Hm Look backwards in the debug log that it dumped out.
Did you see: Net::SMTP>>> STARTTLS
Net::SMTP<<< 220 2.0.0 Ready to start TLS
[02:02]
ssinfodyes
Can I paste the big output here ?
[02:03]
gac410NO
Please don't
let me search a bit.
[02:03]
ssinfodI could use a "paste bin" maybe.. [02:04]
gac410https://support.google.com/a/answer/6260879?hl=en
No i don't need to see it I don't think
[02:04]
ssinfodIs it because it is " Net::SMTP" instead of " Net::SMTP (SSL)?) [02:06]
gac410no, It was "trying TLS" according to your log. hm Yes please pastebin your log.
The email wizard doesn't use the settings. It tries to guess them. Starts with the most secure (TLS) and works down to unauthenticated. From there it will auto-configure the mailprogram setting.
[02:06]
ssinfodhttp://pastebin.com/saq8ewCg
Can you see the paste bin ?
it expires in 1 day.
[02:08]
gac410yes. In the response from gmail is the reason.
https://support.google.com/mail/answer/14257?rd=1
[02:09]
ssinfodbad password ?
or 2-steps verification ?
[02:12]
gac410As the page says under Troubleshoot sign-in problems .. "If you are sure your password is right," then it gives you stuff to try.
Set up an app password, or allow less secure apps. TBH I really don't know how your gmail / google accounts are configured.
[02:13]
ssinfodI would set very standard !.. (default..) [02:15]
gac410Autoconfigure *definitely* works with gmail. I just did it for my account. So you need to adjust your google / gmail account settings to allow it to work. My account is set to allow insecure apps. [02:15]
ssinfodok, I will try this..
it seems to be the reason..
[02:15]
gac410Its was the default when I got my gmail account years ago. [02:16]
ssinfodAnyhow, why foswiki is considered unsecured ? [02:16]
gac410Foswiki is using Secure SMTP using STARTTLS which is the most secure and recommended email connection. Google doesn't think passwords as a technology is secure.
Of course it's considered insecure from an "application" perspective, because you have to put your gmail account password into your LocalSite.cfg.
So if you enable 2-step verification. (Google will send a one-time pincode to your cell phone when you log in) that doesn't work with apps like Foswiki or Thunderbird. So you can give them unique "app passwords" that allows that app to bypass the 2-step.
I've never tried any of this stuff.
[02:17]
ssinfodok.. I will just disable 2-step.
(at least for test)
anyway, this email is a dump..
[02:22]
gac410If you have 2-step, see https://support.google.com/accounts/answer/185833 for how to get an app password. [02:22]
ssinfodI was reading this faq yesterday...
http://foswiki.org/Support/Faq69
It is saying: "Make sure your LocalSite.cfg is not world readable"
[02:23]
gac410Yeah We need to update that to cover the new Gmail 2-factor and app password stuff. [02:24]
ssinfodWhat permission would it be ? [02:24]
gac410Right. You want LocalSite.cfg to have 600. We *should* create it with that permission by default.
ls -la should show -rw-------
Still, I do agree with google on that point. Putting your gmail credentials on a server is not the best practice. But I've not had the energy to turn on 2-factor and deal with the headaches.
[02:25]
ssinfodWhere is LocalSite.cfg ?
is it in /var/www/foswik ?
sorry for all my questions..
[02:28]
gac410in the lib directory. probably /var/www/foswiki/lib
No problem. Email is pretty complicated unfortunately. The 2.0 wizard tried to take the pain away, and then gmail goes and adds more pain. :(
[02:28]
ssinfod-rw------- 1 www-data www-data 31388 Feb 9 20:59 LocalSite.cfg [02:30]
gac410yup. So it can only be read by the apache server. And hopefully you've followed our guidance (either example configs, or ApacheConfigGenerator, so that the file cannot be fished from a url. [02:31]
ssinfodyes I used the ConfigGenerator.. [02:32]
gac410Okay, so lib should be blocked from web access. good. [02:32]
ssinfodOk, I will look around to disable the 2-factor..
thanks for your support.
I think it saves me a lot of search..
[02:33]
gac410Sure. Either that or try to generate an app password. That would be preferred. ;) [02:33]
ssinfodyes I will also try that
thanks again.
have a good evening..
[02:34]
gac410you're welcome
you too
[02:34]
ssinfodsee you
later maybe !!
[02:34]
gac410;) [02:34]
..................................... (idle for 3h0mn)
***gac410 has left [05:34]
foswiki_irc5Greetings.
foswiki.org's Tasks web is based on BugsContrib. foswiki.org sends out email reminders for the Tasks web. Is that feature part of BugsContrib?
[05:41]
......... (idle for 40mn)
EisNerdmoin
someone here who could explain me howto get a debugger attached to foswiki running as cgi?
I have trouble finding a usefull frontend for remote debugging perl
[06:23]
................. (idle for 1h24mn)
***ChanServ sets mode: +o MichaelDaum [07:48]
MichaelDaumgood morning guys
did George already say when he is creating a Release02x10 branch?
time to move forward on master
[08:01]
....... (idle for 30mn)
msnIs there a chance that SignaturePlugin would work [08:32]
................ (idle for 1h16mn)
EisNerdhi I managed to get ptkdb running, but the break button doesn't work [09:48]
LavrHi guys. Finally got the new hardware for the office and setting up Foswiki 2.1. Question. Where are the new macros like PERMSET_VIEW documented? [09:48]
EisNerdso I can't let it run and break after awhile to see where it goes [09:48]
LavrI cannot figure out the magic behind a setting such as * Local PERMSET_VIEW = registeredusers and I cannot find any docu in System web. Did someone forget to document the new access rights Macros? [09:50]
I cannot even find it when I search for PERMSET_VIEW on foswiki.org Development web.
I am trying to figure out if I should hide the Permission tab ot alter its content. I cannot allow our UI to have a one click to allow anyone access because by default all pages are limited to registered users.
Anyone can set an ALLOWTOPICVIEW manually but few does because it is geeky. But "one click" and people will start doing it all the time.
[09:57]
And you can actually with one click shut yourself out from your own page by selecting "nobody" in the permissions tab. That is the most silly trap I have ever seen in a UI. That should not be there!
Unless you are an admin.
[10:13]
MichaelDaumLavr, hi [10:13]
Lavrit should be conditionally there [10:13]
MichaelDaumlong time no see [10:13]
LavrHi Michael [10:13]
MichaelDaumhope you are doing fine. with regards to your question: PERMSET_VIEW is not a macro.
it is an internal variable used by NatEdit to handle the Permissions wiki app in edit mode
[10:14]
LavrTerminology. Preference then. Should still be documented somewhere so people that want to know what the setting is can find it. [10:14]
MichaelDaumit is an implementation detail not meant to be set by anything else but the Permissions wiki app
and it may even change in the future
[10:15]
LavrYes. And that is why you should document it as an internal setting so people know not to remove it when they are editing preferences in More Topic Actions
Just a oneliner in the NatEditPlugin so you can find it with a SEARCH in System web would have done it for me.
[10:16]
MichaelDaumas you are just there atm: could you have a look at the ongoing discussion at http://foswiki.org/Development/ImproveOOModel please
it is important to keep up the discussion there
we have a new high-potential :) contributor doing a lot of work in that area and I'd rather not see all these efforts being wasted
you raised concerns there and as far as I can see we have means to work around them
I'd much appreciate your input there. thanks in advance.
[10:17]
LavrI can see the discussion has moved to something else since I raised the concern. My concern was related to using CPAN libs that are not in the main distributions. [10:19]
MichaelDaumy
see the number of checkins at http://foswiki.org/Tasks/Item13897 ... awesome isn't it
[10:20]
LavrFor RedHat family the CPAN libs should be as RPMs in either the standard default repository or the EPEL repository. These are official and conflicts are very rare when you stick to them. [10:21]
MichaelDaumwell [10:21]
LavrWhen I installed Foswiki now from scratch I ran into 3 CPAN libs that we refer to that were nowhere in the libs. [10:21]
MichaelDaumand this will probably increase [10:21]
LavrThe EPEL contains almost anything from CPAN so it is difficult to find one that is not [10:22]
MichaelDaumi see
did you see CpanContrib?
[10:22]
LavrYou choose something very exotic when it is not there. We can avoid that for sure. [10:22]
MichaelDauma lot of distros either don't ship perl packages that we require or those that are there are outdated, i.e. now that we are all unicodified
Moo is a pure-perl package ... which is just fine to be packaged into a CpanContrib
[10:23]
LavrCpanContrib is sort of OK. But if you read its docu it still requires to build some libs. And that is the problem. The people that have trouble are customer of shared hosting without access to a command line prompt. If you cannot ftp your stuff to the shared host or run it from Apache you are fucked.
Let me check is Moo is in EPEL.
What is the exact CPAN name?
[10:24]
MichaelDaumMoo [10:25]
FoswikiBotMoo is a pure-perl package ... which is just fine to be packaged into a CpanContrib [10:25]
LavrMoo it seems [10:25]
MichaelDaumhttps://metacpan.org/pod/Moo [10:25]
LavrNo package perl-Moo available [10:26]
MichaelDaumMoo is for Perl what ES6 is for JavaScript [10:26]
MichaelTempestHello - foswiki.org appears to be broken for editing. I wanted to fixed a broken (old-style [[link text]]) link on Community/InternetRelayChat [10:27]
LavrWhy it is not in the standard Perl RPM libs then??? EPEL picks up from bleeding edge Fedora world. [10:27]
MichaelDaumas far as I see CpanContrib does not require any additional build stage, does it? http://foswiki.org/Extensions/CpanContrib doesn't say so as far as I trust my eyes. [10:27]
MichaelTempestI get a "confirmation required" when I try to save.
Is it just me (or my proxy), or is something amiss with foswiki.org?
[10:27]
LavrMaybe it is only the Updating of CPAN libs that requires that in CpanContrib. [10:29]
MichaelDaumso once Moo has been added to CpanContrib ... things should be just fine - not only for Redhat/Fedora [10:29]
jastfoswiki.org works for me [10:29]
EisNerdok I got the loop [10:29]
MichaelDaumLavr, instead of installing from EPEL, would you consider isntalling CpanContrib? [10:30]
jastLavr: fedora has it: https://admin.fedoraproject.org/pkgdb/package/rpms/perl-Moo/ [10:30]
LavrI would hate to install CpanContrib to get Moo because it also means that all other CPAN libs in that contrib overrides the system. I really prefer running a production environment on a pure RPM based perl install where the system updated security updates all the time. [10:30]
MichaelDaumLavr, good point [10:30]
LavrMaybe we can trigger the EPEL gatekeeper to pick it up. [10:30]
MichaelTempestjast: Thanks - I see Sandbox works for me. Hmmm. [10:31]
LavrIt may be as simple as asking someone in an email and Moo will be added to EPEL. [10:31]
EisNerdmy FosWIki loops _getMapping => LdapUserMapping::handlesUser => ...userExists => TopicUserMapping::userExists => getWikiName => _getMapping ... [10:32]
MichaelDaumLavr, though things in CpanContrib serve as a fallback in the library path. so any package found on the system library path have higher precedence. in your case only Moo would kick in from CpanContrib [10:32]
LavrEPEL is "Extra Packages for Enterprise Linux" in case someone wonders.
Oh. I though the local would always win over system.
[10:32]
MichaelDaumnope [10:33]
LavrWell. Why don't we try and get Moo added to EPEL anyway? Worst thing is we get a "No" [10:33]
EisNerdit tries to get mapping for WikiGuest [10:34]
MichaelDaumany "use Moo;" looks for (1) <foswiki-dir>/lib/Moo.pm (2) /usr/lib/perl/.../Moo.pm (3) /usr/local/lib/perl/.../Moo.pm and finally as a last resort in <foswiki-dir>/lib/CPAN/lib/Moo.pm [10:34]
LavrI'd better get back to my installation of Foswiki 2.1. I surely need to skin out at least parts of the Preferences tab. Or the whole thing. [10:34]
jastMichaelTempest: any specific URL that doesn't work for you? I only tried the front page [10:35]
LavrI really think Michael that next time you update the NatEdit you make the "nobody" selection conditional to being an admin [10:35]
MichaelDaumLavr, okay. well, in anyway, please remember our irc discussion today and please review http://foswiki.org/Development/ImproveOOModel. This is important! [10:35]
jastI'm not sure I understand fedora's packages site correctly... seems like maybe moo is only in rawhide? [10:36]
LavrI am not sure I understand the ImproveOOModel consequences. IS the idea to fuck up the whole API so no plugins are compatible anymore? [10:36]
MichaelDaumyou bet ;) [10:36]
MichaelTempestjast: http://foswiki.org/Community/InternetRelayChat [10:36]
jastas I understand it, for now the goal is to make internal changes only [10:36]
EisNerdit starts in getCanonicalUserID
someone an idea?
[10:36]
jastMichaelTempest: weird, page loads fine for me, both as guest and logged in [10:37]
LavrWell. Then I guess it will soon be time to fork Foswiki then. Because if you break all Plugins you destroy the whole ecosystem of the project. [10:37]
MichaelTempestjast: The page loads fine. I get "confirmation required" when I try to save the topic. [10:38]
MichaelDaumLavr, we will not break backwards compatibility to a level none of todays plugins wont work anymore.
that is not the point of ImproveOOModel
[10:38]
jastMichaelTempest: oh, right. I didn't see your first two lines, the ones where you mentioned that. :) [10:38]
MichaelTempest:-) [10:38]
MichaelDaumgranted, the main goals of ImprovingOOModel are not clearly stated on the feature proposal topic
such as "current oo model sux"
[10:39]
jastMichaelTempest: okay, tried saving, and I don't get a strikeone message... but I do get a "AntiWikiSpam Intercept"
I also get a few dozen jquery script errors per second while the editor is open
"Syntax error, unrecognized expresison: unsupported pseudo: -webkit-autofill"
[10:40]
LavrI must admit that I'd rather see development effort that improves the user experience than keeping on rewriting the code to do the same. When I look at TWiki they have actually managed to add some cool new things the past years. It is very little that has happened in Foswiki space. [10:41]
msnLavr: i was wondering why i didn't run into the moo issue since i installed all packages from package manager then realised debian :) [10:42]
MichaelDaumLavr, not sure I could agree. but yes you are right in general. [10:42]
jastLavr: I do appreciate core effort, I run into a lot of limitations and subtle breakage in core land [10:42]
MichaelDaumthings that currently suck most are (1) user code (2) search code [10:42]
MichaelTempestjast: ok. Thanks for checking. [10:43]
jastbut some of the issues are huge and would take a LOT of changes. for example, viewing old revisions still shows the newest version(s) of images, metadata rendered by skin templates, ... [10:43]
MichaelDaumi.e. if we tackle (1) would a redesign depend on an already improved oo model [10:43]
LavrThe debian family perl repo is more complete than the RedHat family. Unfortunately it is Redhat family that dominates the "customers" we have. [10:44]
MichaelDaummy main motivation to keep working on the core code - besides any required improvements to foswiki's (lack of) usability - is to make sure the code is sustainable the next 10 years [10:44]
LavrAnd I have burned myself running a mix of RedHat rpms and CPAN installed libs. [10:45]
MichaelTempestjast: Which browser are you using? [10:45]
jastfirefox [10:45]
MichaelTempestMe too. [10:45]
MichaelDaumLavr, Redhat is not the only distro that fucks up perl [10:45]
MichaelTempestMichaelTempest blames the corporate firewall :-) [10:46]
jastI've set up a few wikis with RHEL 5
lots of CPAN there
[10:46]
LavrRedHat does not fuck up perl. It is when you mix the RPM and CPAN. And actually it is not perl that fucks up. it is the RPM system that stops working because of conflicts [10:46]
jastfor added fun the machine is isolated from the internet
RPM shouldn't run into issues due to CPAN installs
so long as cpan is configured to use e.g. /usr/local as its prefix (which is the default on Debian at least)
[10:46]
MichaelDaumwhen a distro ships a perl not supported upstream and then compiles it in a way it is 10-20 times slower than when you compile it locally (via plenv or perlbrew) then - sorry - the distro fucked it up. and thats the case for ALL distros atm, including ubuntu and redhat and whatever is out there. sad story. [10:48]
LavrCPAN is not installing in /usr/local unless you are a geek and know how to set it up. I don't. [10:48]
jastthen the CPAN configuration in redhat seems... suboptimal [10:48]
LavrAnd you all have to remember that the average sys admin knows nothing about Perl and CPAN. They just want to install some software. [10:49]
jastof course in Debian you also have cpan2deb to generate packages yourself
maybe something like that exists for RH/fedora, too?
[10:49]
LavrYou can do everything in this world if you know how to. But do not expect your admins to be programmers! [10:50]
jastgenerating a package in debian is literally one line on the shell
cpan2deb Moo
[10:50]
MichaelDaumfrankly, with perl we are in a very bad situation atm.
Moo or not Moo
[10:50]
jastdoesn't take programming knowledge, you just need the tool(s)
seems like fedora has a tool for that, too, only it doesn't do all of the work for you
[10:50]
Lavrjast - and how does that Moo thing get updated automatically when a security issue is found? Answer it does not!. The advantage of having a clean RPM environment with yum running regularly is that security updates are being installed without much human intervention. In an Enterprise world this is key important. [10:51]
jastI wasn't advocating anything as a silver bullet
realistically if you use RHEL you're stuck in the past and any way to deal with that will have drawbacks, that's just the way it is
[10:52]
MichaelDaumLavr, hehe. That's how I thought too for years. But hey, look at the shipped RPM versions. mostly outdated by design: it is an Enterprise Linux. [10:53]
LavrThe Foswiki 2.1 that I installed now runs 100% on RPMs. There were 3 CPAN libs that I could not find and they are optional and I do not think I need them anyway. Adding Moo breaks this completely for the RedHat world as I see it. [10:53]
MichaelDaumanyway. we have a way around this to some degree for now. [10:54]
Lavrnever forget who the target group for TWiki and Foswiki are. It is not a toy. It is not a personal website package. It is an Enterprise business application!
What I consider hot burning issue right now is that our flagship drawing plugin JHotDrawPlugin is not being updated to not depend on Java. We are down to only having this supported by an outdated Internet Explorer.
[10:54]
jastthat's something that would take several person months of work
and for me personally other issues are more important
[10:58]
Lavryes. But unlike the OO stuff this would actually be important to the users. My users do not give a fuck about what OO model Foswiki runs under. I am not saying not to improve the core. But if that is all we do this project becomes irrelevant very soon. [10:59]
jastand "several person months" was a pretty generous estimate
s/generous/optimistic/
we built our own plugin for this but it's based on a comprehensive proprietary library so we can't distribute it... even with that library we had to do a LOT of work
[10:59]
LavrThe OO stuff will take 2-3 years before it gets stable. [11:00]
I am fighting at the moment to keep Foswiki as the application for our business ISO9000 system and the JHotDrawPlugin is an integrated part for making process flow charts (simple boxes with arrows and links). It is a major threat for me that Java support is removed from all browsers. That is what worries me as an admin. Not how the OO works in the core. I could not care less to be honest.
And for the record. I am a hardware engineer - not a software programmer. I can hack up a plugin here and there and it takes me forever. And I do not know java at all. Otherwise I would be working on it right now.
I have looked but it is beyond my skill level :-(
[11:05]
jastand it's beyond my time level [11:08]
LavrI will try to dig out some interest with our large software pool of people in Krakow, Poland. Some of them rely of Foswiki heavily so I will see if I can get someone to help. They are 1000 software engineer there
engineers
MichaelDaum - I will raise an Enhancement Item to NatEditPlugin to get that nobody setting made invisible unless you are an admin yourself. I hope you agree to change that next time you change NatEdit. I think the admins will appreciate not having too many "help, I have locked myself out of my own topic" requests.
[11:09]
http://foswiki.org/Tasks/Item13951
I like the enhancement you guys did to the macro syntax. The + feature. That will be cool to work with! That is an enduser enhancement that kicks butt.
[11:18]
msni am trying to use foswiki for Change management system its working only edit row plugin doesn't handle 2 people editing same line very well
and the Signature plugin is broken :( that can add an extra "he signed it for change" on the page
[11:26]
LavrNever used the signature plugin. I always used the CommentPlugin to tailor a simple way to add a signature at the bottom.
I have little experience with EditRowPlugin in a live installation but I have already decided to enable EditTablePlugin and rename the macro for EditRowPlugin to EDITROW. I find ERP to be a little too easy to accidently click and fuck up a table cell.
The two plugins work fine both enabled. Just - don't use them both on the same page!
[11:29]
ark987Hello guys its me again :)
Im still trying to figure out the puzzle of the topic display titles with NatSkin. I just installed from scratch a clean Foswiki site, then installed and activated NatSkin. In the next lines Ill explain what I'm doing and what I expect to happen
So from Main page I click New (the pop-up window appears), I type the Title in human readable format: "How to install Foswiki"
I type some contents and the document, click Save.
Go back to Main page, click Edit. Add the magic wikiword that was created for my new page: "HowToInstallFoswiki", click Save.
When I return to the view mode of the Main page I see the wikiword: "HowToInstallFoswiki"
I was expecting to see the "human readable" topic title that I've set earlier using the New button: "How to install Foswiki"
Because of this behaviour, I have to instruct other people to go into Topic Preferences and set the variable: * Local TOPICTITLE = How to install Foswiki
After setting this variable, the wikiword is redered with the human readable title in the Main page as expected.
My question is, Is this the expected behaviour redering topic tittle?
I also came across with this: http://foswiki.org/Development/TopicDisplayName still scratching my head wondering if is implemented, or should I open a ticket...
[11:31]
msnLavr: for me there are not enough people to actually go and edit those but my issue is 2 people editing same row in edit row one, whoever edits first saves his data other person loses his data completly [11:36]
MichaelDaummsn, data loss - this is a show-stopper. [11:38]
msnit is
2 users cannot edit the same row
the merging doesn't work
[11:38]
MichaelDaumomg [11:38]
msnwell they are not editing the same row, they are both creating a row at the same time
editrow assumes its the same row and boom
for now i have tried to build a work around it
[11:38]
MichaelDaumLavr, "I could not care less to be honest."
we probably will have a community vote on ImproveOOModel
[11:39]
msnby splitting the table based on some parameter to make 1 table into 6 so mostly they will editing one row at atime and different tables [11:40]
MichaelDaumworking on the OO-Model does not mean work on other frontiers will halt (putting aside redesign of the user code which pretty much depends on the rest of the core being OO-ified or not)
msn, best is not to rely on tables too much to store data in a robust way ... which it isn't.
[11:41]
EisNerdas I could reproduce mit LdapContrib trouble now in debugger, is someone here willing to have a look on it? [11:44]
msnMichaelDaum: well they work fine as long as there are way to avoid conflicts [11:45]
EisNerdit looks to me that it starts an endless recursion while trying to resolv WikiGuest [11:45]
..... (idle for 21mn)
ark987Dear lord, I finally found a workaround: http://foswiki.org/Tasks/Item13835
For the record what Ive described above is hapening in the latest version: Foswiki-2.1.0, Plugin API version 2.4
After I applied the patch the Topic Title is displayed and the variable is set as expected
[12:06]
LavrMichaelDaum. You will not find me trying to block development of an OO model. I am concerned about Moo. And I am concerned that there is no description of how we will handle compatibility for existing plugin of which many are not public. E.g. I have a plugin that can convert a Risk Excel sheet to HTML. I have not contributed that because noone else in the world has that exact Excel sheet (lucky you). So the OO
page needs to be extended with the strategy for compatibility and how to ensure that upgrading Foswiki does not become a new nightmare. I will add that to the page.
[12:19]
........... (idle for 51mn)
jastEisNerd: I don't have time right now but I'll keep it in mind for the next time I look at this issue [13:10]
JulianLevensWhat creates files with igp_ prefixes? [13:13]
jastImagePlugin
thumbnails and such
[13:13]
JulianLevensty [13:15]
***ChanServ sets mode: +o Lynnwood [13:22]
.... (idle for 18mn)
JulianLevensMichaelDaum SolrPlugin is working on our new Foswiki, but I'd like to ask a few questions:
I'd like to have the old searches available: how would you suggest I go about that?
With the users search. I use to have various facets available to select 'company, location etc'
Well I only have location now: I cleaned up all the user forms to be consistent and changed the {Ldap}{PersonAttribures} to map to the relevant field names in the UserForm and done a full reindex
So, what did I miss?
[13:40]
***ChanServ sets mode: +o gac410 [13:46]
ark987Hey gac410, about yesterday with my issue with the topic titles. You were right, what you suggested to do with natedit.tmpl is the way to go. I also found this which is exactly my problem: http://foswiki.org/Tasks/Item13629
Oops wrong link , I meant this one: http://foswiki.org/Tasks/Item13835 I suppose this hole Topic Tittle thing is a work in progress
[13:57]
gac410Lavr: Kenneth, rather than just raging on about choice of Moo when *Every Other significant distribution includes it* Have you considered moving your complaints over to an EPEL support forum and asking them to include it?
It may be that nobody has asked for it, Or maybe they have a good reason to Not include it, which would be justification for us to reconsider the decision.
[14:04]
MichaelDaumhi gac410 [14:08]
gac410jast: regarding branching 2.1, Do you have to do anything special on weblate or can I just create the branch project.
Hi MichaelDaum
[14:08]
MichaelDaumwas going to as about it: when do you consider creating a 2.1er branch [14:09]
gac410Any time. BTW, I can always branch at a commit instead of the HEAD, so it's not a disaster if you have stuff to do. [14:10]
MichaelDaumJulianLevens, to get back the old searches just switch off the related {AutoTemplatePlugin}{ViewTemplateRules} [14:10]
JulianLevensSorry, I want both [14:10]
gac410I just figured if Jast was around I could coordinate the branch with the weblate work. [14:11]
MichaelDaumfor example have 'Wearch' => 'SolrSearchView' but leave 'WebSearchAdvanced' untouched
so you will have WebSearch solr-ified while WebSearchAdvanced remains untouched
of course a http://.../WebSearch?template=view will give you back the non-solr version on an adhoc base
or edit its preference settings and add a manual VIEW_TEMPLATE = view to it to make that choice permanent on a per web base
the normal view template drill
[14:11]
gac410MichaelDaum: Regarding onclick vs. auth. I considered that. It seems that there are legitimate times a guest might want to refresh a cache. Just blocking it altogether is rather draconian. [14:13]
MichaelDaumwhen do you think would a guest care abouit the page cache? [14:14]
JulianLevensThanks, the 1st option works and that's good for now, I'll remember the options for future reference [14:14]
gac410Looking for the latest info. Even running my own server with default cache settings, I've needed to reset as guest.
Depends on how thoroughly a tuning job the admin has done
[14:14]
ueberallMichaelDaum: Whenever I look at a page which carries information I'm interested in, I look at dates. If there is a refresh option, I always click on that first in order not to waste my time reading outdated material, even if we're talking about an hour since the page was created. [14:15]
MichaelDaumhow comes the info is outdated and you'd refresh manually?
all pages are auto-purged by the foswiki core
[14:15]
gac410I don't know. Page with a search that is cached? [14:16]
ueberallIf there is no chance that the generated page can be outdated, the notice should be removed altogether. [14:16]
MichaelDaumueberall, true [14:17]
gac410I've definitely see outdated pages. [14:17]
MichaelDaumthe only case when cached pages are outdated is when something changed under the hood
changes to the txt on the filesystem by a cron job or so
[14:18]
gac410What about missing dirty area tags. [14:18]
MichaelDauma couple of pages are well known to be difficult to cache and keep fresh but also have the most value from being cached ... such as WebSearch and WebRss [14:19]
ueberallWhenever there are dirty area tags, the notice makes sense (no easy way to rule out changes). If there are not dirty areas/queries--e.g., changes to the filesystem, this could be dealt with by watching those changes and automatically recreating the pages in the background. [14:20]
MichaelDaumbut these are listed in $Foswiki::cfg{Cache}{WebDependencies} by default and thus are recomputed by any change in the web no matter whether there is a content dependency or not [14:20]
gac410What is your objection to the on-click. It should be effective for the bots and still permits a guest to refresh a page. An occasional manual refresh isn't all that onerous. [14:21]
MichaelDaumcontent security policies
javascript embeded in html is insecure
[14:21]
gac410We use that all over the place though [14:22]
MichaelDaumonsubmit, onclick and how they are called are all well known for their xss [14:22]
gac410Well I'm not comfortable removing refresh cache for guests. I've had to refresh too many times. [14:23]
MichaelDaumueberall, I think you are right: this caching info is mostly bragging ;)
gac410, up to you. I am just a guy with a different opinion.
[14:23]
gac410I would prefer that WebRss NOT be cached. I use it as my PRIMARY way to see what's going on. [14:24]
EisNerdjast: great [14:24]
MichaelDaumgac410, woops
not caching WebRss?
wtf
[14:24]
gac410I use Firefox live bookmarks. [14:25]
EisNerdjast: if you have some ticket/report I could attach the information I colected [14:25]
MichaelDaumgac410, just to make this clear: WebRss is invalidated on EVERY edit change on the web. so it is always UP2DATE. [14:25]
gac410Oh.... okay. that's good.
never mind then.
[14:25]
MichaelDaumdisabling this feature means: your server is recomputing the same rss feed thousand of times again and again. with each rss reader doing an active poll every now and then, this is THE best way to waste server resource since the invention of atoms [14:26]
gac410Okay I have not disable it. It's foswiki.org where I use the live bookmarks. I was just concerned that maybe I'm missing stuff. [14:27]
MichaelDaumy
;)
[14:27]
gac410I suppose remove the link, and then if a user really needs to get a fresh copy, they could add refresh=cache to the url.
BTW. speaking of URLs. The *URL macros DO automatically append unknown options to the generated URL.
[14:28]
MichaelDaumxcellent. didnt know that. wasnt in the specs or the code.
forgotten to mention anchors
[14:29]
gac410It's right in the docs Foswiki:System.VarSCRIPTURL "any other parameters to the macro will be added as parameters to the URL" [14:31]
FoswikiBothttp://foswiki.org/System.VarSCRIPTURL [ VarSCRIPTURL ] [14:31]
MichaelDaumthings like %SCRIPTURLPATH{"view" topic="Main.WebsSearch" anchor="q=foo&rows=10" [14:31]
JulianLevensstrictly that's fragment not anchor [14:32]
gac410Yeah anchor is missing. And it needs "special processing" to be after all query parameters.
right.
[14:32]
MichaelDaumah ok got it wrt url params
then my recent comments are pretty bogus
Was looking at PUBURL http://foswiki.org/System/VarPUBURL
and it is not mentioned there
[14:33]
gac410I *think* they use the same underlying code ... not sure. Good point to check.
Too bad our macro processor doesn't allow #="blah" as an argument. :(
[14:34]
JulianLevensBTW: our Main web is our primary web not just for users, but that's not really best practice is it? [14:35]
gac410gac410 doesn't like english arguments. as you point out new ones might collide with existing web apps. and 2) they can't be translated. [14:36]
JulianLevensactually #='blah' would work [14:36]
gac410hm. Just tried it before I posted that comment ;) [14:36]
JulianLevensthe regex only says any non-'=' when getting the key
hm myself
[14:36]
gac410ping jast ... you around?
JulianLevens: yeah Main is both users and default home web. HomePagePlugin though can override that.
Though it's documentation leaves one wondering "the web or topic to when none is specified, or on login/logoff"
If you can parse that ... please do :D
[14:38]
JulianLevensOh, it's META attributes that are very unforgiving on the name, MACROs OTOH are [a-zA-Z0-9_] so no # :( [14:42]
gac410But it does work. For instance, Set that to "Main.WebHome" and set {UsersWebName} to "Usersweb" and you have your separation.
JulianLevens: Hm. In the interest of I18N, the macro should be more generous for the %*URL*% Macros. That needs to be fixed.
[14:43]
JulianLevensOk, and why is it better practice? It feels right somehow but why? And indeed why not? [14:45]
gac410Otherwise can't use canonical URL generation with internationalized query strings. [14:45]
Lavrgac410 - I actually did mention earlier today on IRC that we could try and ask for the Moo to be added to EPEL
Moo is my least worry about the proposal.
My worry is the idea of rewriting everything draining energy away from making features that are actually visible to the users
[14:45]
gac410I'm sure (I hope!) we know that there will need to be some compatibility layer. [14:47]
MichaelDaumLavr, just to make this crystal clear: even a Moo-ified Foswiki is supposed to run non-Moo plugins. [14:47]
LavrBut I cannot hold you guys back if that is what you want. I just ask you to THINK and take things in small steps [14:47]
gac410Regarding draining energy. We have a very motivated dev. who is willing and has the time and the itch to scratch, to work on this. [14:47]
MichaelDaumalso note that nothing is holding you back in implementing other happy features [14:48]
gac410It's valuable work, and moving away from the DEPRECATED CGI is somewhat related to all this
PSGI is another effort needed and this is all related.
[14:48]
LavrDeprecated CGI is a concern indeed. [14:49]
gac410BTW ... the other project has chosen to bundle an old CGI rather than fix their code to address the security issues fixed in the latest CGI. [14:49]
LavrI do not think they have chosen that....
... when you have no developers that is a gunshot wedding kind of decision
[14:50]
gac410It was not all that much work to scan our code and change calls to Request->param() to Request->multi_param() to separate out the scalar vs. array type parameters. [14:51]
jastgac410: hi, I'm semi-here [14:52]
gac410Took us a few days of fidding [14:52]
vrurgI'm right on time here. MichaelDaum is right, nobody is gonna drop off support of legacy plugins. [14:52]
jastI can't think of anything special in Weblate, so go ahead and let me know if it breaks ;) [14:52]
gac410Okay great... I'll branch Release02x01 today then, so Developers: Start your engines :D [14:53]
jastbranching on github doesn't affect weblate at all
you have to create a new sub-project (or something like that) on weblate for it to pick up on the branch at all
[14:53]
gac410Right, but I don't want weblate to get out of sync with translations between master and the new 2.1 branch. [14:53]
LavrGetting closer to putting 2.1 in production. Just finished making my tailoring of NatEdit. [14:53]
gac410Lavr, have you seen the discussions of regex performance on old perl. [14:54]
LavrNo [14:54]
MichaelDaumwith regards to NatEdit: I am about to factor out the "engine" part from the framework, so that different once can be plugged in such as codemirror, prosemirror, ckedit or tinymce [14:55]
gac410Regexes are a lot slower before perl 5.20 [14:55]
jastEisNerd: I don't have any issue open yet, but feel free to e-mail me (krueger@modell-aachen.de) or create a Task for LdapContrib on foswiki.org and assign it to me (JanKrueger). I can't promise any specific timeline but will try to look at it in the next week or two. [14:55]
LavrMichaelDaum - the tailoring was "easy" because you may remember you told how to do it some months ago and that all worked spot on so thanks for those hints [14:56]
MichaelDaumyw [14:56]
LavrHmm. My new Centos 7.2 is at perl 5.16.3 so I guess I will not have that advantage yet [14:57]
jasta few of our wiki installs are at perl 5.8.8, so you're basically from the future [14:57]
LavrI cannot say what is new server and what is new Foswiki. But the performance of the new combo is - tada - factor 2
Normal page load goes from 1.3 second to 0.6 second.
[14:57]
gac410excellent. The regex issues show up on big pages. MichaelDaum Was it big search results you saw the slowdown?
We switched foswiki.org over to plenv to get the latest. But not for a specific performance issue, just general concern over MichaelDaum's findings.
[14:58]
MichaelDaumWebTopicList
slowdown was in regexing $perce?nt -> %
decodeFormatTokens() took up to a third of the overall processing time ... very strange.
found out via nyt profiling
same page on 1.1.9, same machine -> no such problem
as soon as I switched to a 5.22.1 perl problems just vanished
conclusion: regexing unicode strings was optimized further more between those perl releases
[15:00]
gac410I found the same issue with EditRowPlugin, but the issue was due to html entities escaping all the quotes. Changed to single quotes, to avoid the escapes, and problem went away [15:02]
LavrOdd. I do not see any performance issue on a WebTopicList with a web with 1400 topics. It loads in 1.04 second [15:02]
MichaelDaumgac410, yes this fits into the picture. [15:03]
gac410fast CPU I guess Lavr :D [15:03]
LavrYes. It is a new fast 6 core XEON
And 32 gigs of RAM
And two 10000 rpm SAS drives in a raid 1.
[15:03]
MichaelDaumLavr, the difference on my system was: WebTopicList loading 1s on 1.1.9 ... versus >10s on 2.0.3 [15:04]
LavrAnd fans that makes more noise than the Spaceshuttle at takeoff [15:04]
MichaelDaum2116 topics [15:04]
LavrI wonder if there are Perl versions in between that had the issue. Or maybe the RedHat perl has backported patches.
Which perl version was slow for you?
[15:05]
gac410I think we dug into the performance and it was not patches. Some of the regex parser was rewritten I don't remember now, but regexes have a slow path and a fast path, and unicode was dropping the parser into the slow path IIRC [15:06]
MichaelDaum5.18.2
stock ubuntu
[15:06]
LavrBy the way. The 2.1 seems much more robust to none UTF-8 chars in topics. [15:06]
gac410Don't follow that one. [15:06]
LavrThe 2.0 was crashing ugly. The 2.1 just shows the none-utf8 as \xx [15:06]
gac410Right. we worked on that a bit. [15:07]
LavrI did a test run yesterday with the convert-charset.pl --encoding=cp-1252 and it went well. I have not done the final one yet. [15:08]
gac4102.1 should also sort better, and be much more compatible with osx clients, or servers [15:08]
LavrI noticed that if you run it again on already converted content it goes wrong. [15:09]
gac410yes. Disaster to run it more that once.
Otherwise it's taking bytes that are utf-8 and trying to make sense of them as cp-1252
[15:09]
LavrSo I need to get everything set and then stop the production server. (or put in read only) and then do the transfer and conversion.
And the hard part is then to apply the changes I have to do to various applications.
I could not select to remove the natedit skin but I like to have the new editor feature. I just need to remove permissions.
And for select applications I need to swap form and text tabs.
I have some apps (similar to our Item bug web) where the real content is in the form and not in the text area
[15:10]
JulianLevenshow do you swap form and text tabs [15:12]
LavrSo then I need the form to be the default instead of text. I do that with a view template on the form definition.
To swap form and text on topics with a specific form you do two things
On the XxxxForm topic you add a %STARTSECTION{"edittemplate"}%SubmissionRecordEdit%ENDSECTION{"viewtemplate"}%
OR if you use the Auto template by name you just create the view template with the right name. That is actually default
And in the view template you have two lines only
%TMPL:INCLUDE{"edit"}%
%TMPL:DEF{"tabs"}%%TMPL:P{"formfields"}%%TMPL:P{"firsttab"}%%TMPL:P{"settingstab"}%%TMPL:P{"helptab"}%%TMPL:END
My example there also removes the permissions tab.
If you still want the permissions...
%TMPL:DEF{"tabs"}%%TMPL:P{"formfields"}%%TMPL:P{"firsttab"}%%TMPL:P{"settingstab"}%%TMPL:P{"permissionstab"}%%TMPL:P{"helptab"}%%TMPL:END%
[15:12]
JulianLevensThanks, trying that now [15:17]
LavrI do not want to do that in general in a skin because on 99% of topics the current default is the right one. Text first and form second. it is only in apps where all content is in forms that I want to swap [15:17]
JulianLevenszactly my situation [15:18]
gac410lavr Someone sent an email to foswiki-infra complaining they could not log in.. After I dug into it, turned out they were trying to use your motion wiki. [15:19]
LavrHa. OK. There are some funny fellows around sometimes. [15:20]
gac410Do you want me to forward you the message? [15:20]
LavrSure.
I just received a message from a bloke that registered and then asked me what his password was.
[15:21]
gac410JoelK ? [15:21]
LavrI do not remember.
Often these same blokes end up just being spammers.
[15:22]
gac410Well if it was JoelK he took the time to respond to several emails back & forth with me... So persistent. Anyway, his initial request it on it's way. [15:23]
LavrOK. I will look at it.
I get so many emails from people asking for support with Motion and I ignore them all now. I have written many places on the wiki that people need to ask questions on the mailing list and not directly.
On the mailing list or the wiki I have tons of people that will help replying and 1000s of people can see the reply. Private emails is a waste of time.
I have started searching for someone to take over the Motion project. I have lost interest of maintaining it.
[15:23]
gac410I used to use it many many years ago. It was nice software. but lost interest [15:26]
MichaelDaumLavr, thanks for your time writing up a clarification at http://foswiki.org/Development/ImproveOOModel
I think your point is clear
[15:27]
Lavr"Go ahead but think while you do it" might have been the short version. [15:28]
MichaelDaumhey good baseline summary [15:28]
Lavr"and don't count on me testing the bugs out because I have no personal interest in the result" [15:28]
MichaelDaumup 2 u as always [15:29]
LavrI wanted to shorten the text I wrote but then I had two meetings.
It takes time to be brief
[15:29]
MichaelDaumwhat I REALLY like about this proposal is that there is somebody (Vadmin actually) that brings in a massive contribution that really is of a lot of value not obvious at the moment but long term. [15:30]
LavrYes. But when I read IRC logs I am not sure there is a clear concensus on compatibility [15:31]
MichaelDaumthink of it as a massive present: birthday + chrismast together, not bought at walmat, HANDCRAFTED [15:32]
vrurgLavr: May i ensure you that I do have clear interest in the compatibility? I do want a new, cleaner plugin model. [15:32]
gac410Lavr What was so difficult about your install this week.
Why was it so painful.
[15:32]
vrurgBut not by throwing in replacement while dumping the old. [15:33]
MichaelDaumLavr, there _is_ a clear consensus on compatibility [15:33]
LavrI know that IRC talk is also about testing assumptions and reading some logs out of context often gives a wrong impression. But what is important is... [15:33]
vrurgvrurg gone for reading Lavr's summary. [15:34]
Lavr100% no nonsense compatibility with topic content [15:34]
MichaelDaum? [15:34]
LavrI think we all agree on not changing names of Macros and syntax etc so that is not so relevant.
For plugins the compatibility needed is that existing plugins should continue working.
[15:35]
MichaelDaumthats the goal [15:35]
gac410BTW.... Plugins that rely on java. Very unlikely unless a motivated user comes along who wants to rewrite. [15:35]
MichaelDaumyea that part of Lavr's summary is probably not relevant to the OO model of Foswiki. [15:36]
Lavror working with a minimum of very well documented changes that the author needs to do like adding an extra statement at the top [15:36]
JulianLevensWhat do you really need from JHotDrawPlugin? [15:36]
LavrWhat I need is two things.
Some way to be able to edit existing JHotDraw drawings in topics. one browser after the other stops working with Java.
[15:36]
gac410The *browsers* have deprecated java as a plugin. So that just isn't going to get fixed by a quick patch. Someone needs to completely rewrite jHotDraw in javascript. [15:37]
MichaelDaumLavr, changes are probably of the kind converting code like $session->{webName} to $session->webName [15:37]
LavrI would not rewrite JHotDraw. [15:37]
gac410And if the 3rd party authors who wrote the java component don't write a javascript version, you are toast. Completely utterly outside of this project. [15:38]
JulianLevensy, I know that there are some really good JS drawing tools out there - much better than jHotDraw [15:38]
LavrI would search for a new existing JS based drawing frame work and and make a new Foswiki integration with that - not compatible with existing JHotDraw.
But then what about the existing JHotDraw stuff? I would hope that we could make a work around where the jar can run stand-alone.
[15:38]
JulianLevensy, hopefully you can convert the existing JHotDraw files to something else, preferably a standard such as svg [15:39]
gac410Lavr, I don't know java, and jhotdraw (the java) is someone elses project altogether. Not a clue. [15:40]
Lavrif you look at the JHotDrawPlugin today it has a perl part which communicates via rest
And the java part which is from an old version of JHotDraw.
If that old version could be altered to a stand alone Java app that I can download and run like a program and it can open a draw file and save it and I can upload that manually... Any dirty work around would do it for me. Just to maintain existing drawings.
But I do not expect anyone to make a JS version of the Java. That would take a year
or two
[15:40]
MichaelDaumLavr, have a look at https://github.com/SVG-Edit/svgedit [15:43]
LavrWhat Foswiki needs going forward is a good kick ass simple vector graphics feature [15:43]
MichaelDaum^^
other options:
https://www.draw.io/
[15:43]
gac410Anyway to change the subject that I can maybe contribute to. What was so painful about a new install. You state "What a PAIN! I have installed all Twikis and Foswikis hundreds of times since Twiki 3.0 (Cairo). And I have never been through such a painful installation." [15:44]
LavrYes. I think I looked that that and I have seen some other cool ones [15:44]
MichaelDaumhttps://www.gliffy.com/
http://creately.com/
http://www.yworks.com/products/yed
http://gojs.net/latest/index.html
http://www.js-graph.com/
https://www.lucidchart.com/
etc
[15:44]
Lavrgac410 - that takes a long answer the PAIN part. But I have taken detailed notes of the installation and I will be writing a document describing the steps (in case I get hit by a bus).
And with that description I can also talk about pain points.
[15:45]
gac410Lavr, we even now include the example "yum" statements needed to install all the dependencies. In SystemRequirements. [15:46]
MichaelDaumjhotdraw is just ... gone out of the window [15:47]
LavrI did not know we had that in SystemRequirements
JHotDraw as OS project is dead. Not worth look at.
[15:47]
gac410yum install -y epel-release
yum install perl-Algorithm-Diff perl-Archive-Tar ...
[15:48]
JulianLevensI am working on improving installation and maintenance [15:49]
gac410Of course since I don't have a rhat system to test with, this depends upon users to feedback issues. [15:49]
LavrBut the pain is mainly the configuration and getting the tailorings to work. So many small things along the way. Just getting email to work took time.
configure has not become simpler and there are some odd things in the work flow.
[15:49]
gac410If server sendmail was functional, email *should* be a single click and save after filling in just one field. - the webmaster email. [15:50]
LavrEmail as an example. The first tab - you set up parts of the email system. And that fails. Until you realize that you must enable it on the 2nd tab. If you walk through tabs left to right, that is really a trap. it took time before I realized that it was a matter of enabling mail on tab 2.
Enable email should be on tab 1
[15:51]
gac410No... Click the auto-configure wizard. If it is successful it enables it for you.
You never want to enable email until the wizard tells you it's working.
[15:52]
LavrIt failed for me
Until I enabled email
[15:52]
gac410That makes no sense. The whole point of the wizard is to find a config that works and then enable email. So you don't block registrations with a broken email system.
Do you use the server email? sendmail, or postfix, or ssmtp or the like, or are you using the Perl Net::SMTP
[15:53]
LavrNet:SMTP with an external SMTP server.
None authenticated. it is inside our firewall
[15:56]
gac410Okay. hm... So your server itself cannot send email? from cron and the like? [15:57]
LavrI may be able to but to ensure the mails are accepted at the receiver I have to use a specific SMTP server
I entered my webmaster email and the smtp host field on the basic settings tab
[15:57]
gac410I guess I'd need to see the results of the wizard filling in just the server name. to see where it went wrong. It should have been able to discover the connection. [15:59]
LavrThe configure script is often sort of hanging. Just now I tried to disable the mail and submit. And the progress "ring" just keeps spinning forever.
I tried autoconfigure now and it went OK. And enabled the mail it seems
[16:00]
gac410Okay. My guess is that it's timing out on unsupported protocols on your email server. We start with modern client standards. We try SUBMISSION port first, with STARTTLS then SSL, finally plain old port 25. [16:02]
LavrIt is the validation screen that triggers the eternal progress mouse pointer. [16:02]
gac410Ah... What browser? [16:02]
LavrYou think it is working but it is done and you can click Done. I am in Google Chrome. [16:03]
gac410okay. yeah. We have a css issue on chrome. I don't understand it well enough to fix it. Firefox doesn't put up the spinner. [16:03]
LavrIt is also confusing that there are 3 auto config buttons on the email setup.
Are they the same
?
[16:04]
gac410Yes
The userid / password fields don't even show if you just put in the email address And some servers don't need user/pass
It's unfortunately part of the architecture of configure. I don't think we have any way to put a button on that isn't associated with an input field.
[16:04]
LavrHmm. I can see there was good help text hidden behind the ...more. I think configure uses the hiding of help too much. You click yourself to death on ...more and most of them reveal 3 more words. I never clicked the top one.
By the way - the CPAN libs I did not install were related to signing of emails. strange CPAN libs that are not in the RedHat repos
[16:06]
gac410yeah signing of email is pretty arcane. They are completely optional. [16:08]
LavrCrypt::SMIME, Crypt::X509, Locale::Maketext::Lexicon, Locale::Msgfmt are not there [16:09]
gac410we had a contributor .. for a while ... who was very very concerned about email spoofing and felt every SSL connection should be validated with it's CA, and the every email should be signed. [16:10]
LavrI found these by running the tool tools/dependencies
we used to have this feature in configure. That was easy to use. it showed what you missed and which versions you had
it is gone
[16:10]
gac410System/PerlDependencyReport
And System/FoswikiServerInformation
[16:11]
Lavrif you know about [16:11]
gac410I think it's all documented. But yes it's new. [16:12]
Lavrand hidden well [16:13]
gac410With the new configure architecture, some simple stuff became very difficult to do. Rather than complicate the heck out of configure, general reports, etc. were moved over to the System web.
But at the same time. The ability to fully check settings BEFORE saving them will avoid some nasty outages that the old configure would happily allow.
gac410 hates to remember the number of times he's taken down a site due to a typo in a regex, etc. in configure.
Most of the verbiage on the configure screen is actually rendered in a javascript TML processor. Everything goes to/from server in JSON format, and the javascript renders it.
[16:14]
LavrYes. Another issue I had in configure is that is guesses all the paths incorrectly using a .. in the path. I think I reported that a long time ago. I had to manually edit that many many places
That guessing of paths used to work fine.
It gave me some trouble I remember having the .. in the paths when I had multiple versions symlinked. So I had to edit that.
I cannot remember what broke.
[16:17]
vrurgLavr: I tried to explain the point in http://foswiki.org/Development/ImproveOOModel if you're still interested. :) [16:21]
gac410Hm. There is some condition in the bootstrap process that does use the ../ style paths. I don't remember now why that had to be.
The comment in the code says "# Note that we don't resolve x/../y to y, as this might confuse soft links"
[16:21]
Lavrvrurg. A statement like $session->{user} replaced with $session->user. If I have to do changes like that to make existing plugins work. But I guess that is not really the case. When it comes to current API. The Func calls are all pretty clean and easy to use. The Meta part is more hocus-pocus and I so not trust touching anything in a session object which in my view is not stable API today
Anyway. In my daily life the trouble I encounter is mainly
People copy paste into the Wysiwyg editor and the result is often garbage when they copy from something very advanced (Microsoft crap)
I'd like to see the Wysiwyg more robust. Not new features. Just robust of what we have
If you do one little mistake in the editor a TML table is saved as HTML and hell is loose.
[16:26]
gac410Hm Lavr, I'd love to see your bootstrap log. I just re-tested bootstrap on 2.1, and did not get any relative paths. Not sure why it's generating /../ paths for you. I'm on plain old Ubuntu apache [16:29]
vrurgLavr: Foswiki::Func will remain in place though I'd like to see it deprecated. Another point of developing a new design is to get nice and clean OO API. [16:29]
FoswikiBothttp://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Func [16:29]
vrurgAnd how much of Perl is in Wysiwyg? I guess most of it problems are in JS code. [16:30]
Lavrit is both JS and Perl. [16:30]
gac410Lavr, we are rather stuck with TinyMCE. They release a new version that pretty thoroughly broke backwards compatibility. I tried to update once but our exits were non-functional.
And I suspect it's getting harder. As HTML moves to HTML5, the wysiwyg conversion back to TML gets harder, or at least changes a lot.
[16:30]
LavrYes maybe. But none the less. What we have often puts people in trouble. I have seen well working SEARCHes and tables garbage up by one unlucky edit/save cycle by the Wysiwyg editor.
And what we really need is a Wysiwyg way to adjust table column widths without turning the table into HTML. ie by setting the right values in a TABLE macro
[16:32]
gac410we really need developers to work on this stuff. Only so many hours [16:33]
LavrWe have many apps that picks info from TML tables and they break to pieces when a table is saved as HTML
Example
[16:33]
gac410yup. it's a royal pain I understand the issue. [16:33]
Lavr| _Milestone 1_ | 20 Sep 2016 |
Some search that goes through a topic and picks up all rows where the text is _italic_
Once saved as html you can guess how that SEARCH is going to work :-)
And having to cut and try and cut and try to get the right number of pixels in TABLE columnwidths="240,50,50" working is also a pain. That is the kind of features that the end users would give you blowjobs for
(and taking the teeth out first)
[16:34]
gac410Lavr, part of the restructuring to Moo, PSGI, etc. is to hopefully attract new developers who want to work with newer stuff. and make it clearer for the new person to understand wtf the core is doing. Will it help. No idea really. But one hopes.
the move to github, totally inconsequential for users, seems to have helped a small bit. We occasionally do get a pull request out of the blue.
[16:38]
Lavrthat is good [16:40]
gac410The Unicode push, again inconsequential for many especially here in the US, hopefully encourages contributions from users where i18n is critical. [16:40]
LavrThe utf8/unicode is a pain for us upgraders. But needed to do proper none English working sites [16:41]
gac410I suspect vrurg's involvement came partly due to our improved support for unicode and utf-8.
jomo as well, who was instrumental in getting the unicode, and later osx support working.
[16:41]
LavrIt is good to have new developers for sure. But it is also important for me to repeat that for the endusers there are some pain-points that have been untouched for years.
And new ones that come.
[16:42]
gac410true. But we either move forward, or stagnate, and die. [16:43]
LavrLike loosing the ability to make drawings in Foswiki [16:43]
***BenjaminMartin has left [16:43]
vrurggac410: Absolutely correct. In a tri-lingual environment utf is what I was praying for a long time. [16:43]
gac410we've got 300 some odd plugins that really need dev's to work on. the small core of dev's cannot do it.
JHotDraw is just one of many, unfortunately it's a much more difficult one to address.
[16:44]
LavrAnother that needs attention and where we can pick from TWiki - is the ChartPlugin. But that is a task I have the skillset to do so I plan to do that soon.
ChartPlugin is cool but it has some bugs and some lacks. But cool it is.
[16:46]
gac410And we are in for a world of hurt, (but are *much* better off than the other project) as sites move to perl 5.22, lots of old code stops working. Perl deprecated unescaped braces in regexes. we are testing on 5.23 as well as 5.8.8 [16:46]
LynnwoodGreetings all. Anyone around use ControlWikiWordPlugin at all? And do you experience it inconsistently working? On the one site where I use it, it only seems to work maybe half the time. (E.g. about half the time loading pages, words in the controled list try to link anyways). [16:46]
vrurgvrurg wants a user-level MAKETEXT support. Who would take care of that?? [16:47]
LynnwoodPersonally i abandoned ChartPlugin for other charting options. [16:47]
LavrYou have something better to share?
I can use some ideas. We all need to show metrics from data
[16:47]
LynnwoodIt's been a while... let me look. [16:48]
gac410ugh. vrurg, that is tough. MAKETEXT needs backend support - more of a developers tool.
Lynnwood: controlWikiWordPlugin. we use it on Foswiki.org. I've not heard of issues.
[16:48]
LynnwoodOn charting, here's one I've used that's very nice: http://www.highcharts.com/ [16:50]
vrurggac410: I know. A plugin with another macro name would be good enough for me. But still there must be somebody to take care of it. :) [16:50]
gac410Yikes. And it's not working on foswiki.org in the example. [16:50]
LavrLynnwood. I will look now [16:50]
LynnwoodFree for non-commercial use. Or iir, license for me as single developer to use on multiple sites was very reasonable.
and i found it easy to shift from chart plugin. I just needed to reformat data to chart.
[16:51]
LavrI would need multiple developer type of license. I really try to keep my site 100% free OSS. It is one of the killer argument why they allow me to continue.
"it does not cost us anything. It is free!"
[16:52]
LynnwoodOne feature that clients really liked was ability to download images of charts.
Lavr - yes i agree. I only used this one after looking at other options and this was the only one that address all needs of client.
[16:52]
gac410Lynnwood: Strange. On first view Foswiki:System.ControlWikiWordPlugin the stop links didn't work. But I've refreshed the cached page a dozen times now and it's worked ever sense. [16:53]
Lavrit looks very cool I must admit. [16:53]
FoswikiBothttp://foswiki.org/System.ControlWikiWordPlugin [ ControlWikiWordPlugin ] [16:53]
LynnwoodThe particular feature we needed that this one provided was "confidence intervals" [16:53]
gac410Don't know Lynnwood ... I can't recreate the fail now. :( [16:54]
LynnwoodThis was several years ago now. I haven't searched recently to see if there is a newer FOS product available.
gac410 we're not using caching, but we are using fcgid.
[16:54]
gac410Yeah It must be something with persistent perl. Just failed again, and then I went to check and it worked. [16:55]
LynnwoodI have monitored page loads while looking at server log and see no errors when it doesn't work, so apparently it _thinks_ it's working even if it's not.
I have also tried changing the load order - putting ControlWikiWordPlugin toward beginning but that doesn't seem to help.
[16:55]
gac410I think I see the bug. [16:56]
LynnwoodLavr - the other thing i'll say about Highcharts is that it's one of the highest quality products I've used. e.g. just works, docs are very good, etc. [16:57]
gac410If topic contains a NOAUTOLINK for the whole topic, it sets a shared variable $disabled=1; It needs an else $disabled=0
Around line 66
[16:57]
LavrLynnwood. Yes. But looking at the license conditions it seems - from the way we use it - I would need an enterprice license for it. [16:58]
Lynnwood...which is probably pricy
they make their money somewhere... and it's not from me. ;-)
[16:58]
LavrThe individual license at 150 is low. So it may be fair. A couple of thousand maybe.
I have seen a lot of stuff lately that is free. But it still needs to get integrated in a plugin with a good macro syntax.
If it gets too geeky only I will use it
By the way. If any of you do nested searches that are slow as hell because they are repeated too many times make sure you check out MultiSearchPlugin
That is the best I have produced for our wiki the past year. The performance boost in some of our applications is like night and day
[16:58]
gac410Lynnwood: If you can try to see if a patch, insert right at the top of the ControlWikiWordPlugin::init() method. $disabled=0; That should init the shared variable. [17:01]
Lynnwoodgac410 - thanks. I've give it a try and report results. [17:01]
gac410If it works, could you open a task? I'll get it fixed. ... or if you feel like maintaining a plugin ... go ahead :D [17:02]
LynnwoodLavr - I've been wanting to test MultiSearchPlugin because I'm sure it would be a big help to me in a number of situations. I regularly have to do very complex nested searches. [17:03]
LavrI have a presentation where I use it twice. And it loads in 5 seconds presenting results from 1400 records. With normal nested search it would take between 1-2 minutes!
Instead of searching the same data many times, it picks up the data once, place it in hash tables, and then does the searches in the tables in RAM. Makes all the difference.
But it cannot do any nesting. It is good where you have to search for one set and for each returned value you need to perform a new search.
Especially the interval mode where you have to count or present data based on a set of intervals of a form field value.
[17:04]
Lynnwoodgac410 I'm assuming the code you're referring to is in ControlWikiWordPlugin.pm. Can you say about what line? [17:08]
LavrCounting the number of bugs raised per week for a year is 52 searches. And that may even be 2 or 3 times per interval to count open and closed bugs. [17:09]
Lynnwoodright after line that says "sub initPlugin {"? [17:09]
gac410Lynnwood: you could add it in line 47 [17:09]
LavrThe MultiSearchPlugin does it in 1 search. instead of 104. That is why it is so fast. [17:09]
gac410After the #my ... [17:09]
Lynnwoodgot it [17:10]
gac410The bug is that it's a global variable, and once it's set, nothing resets it.
So initialize it to zero, should hopefully fix it.
[17:10]
Lynnwoodgac410 - it seems to work initially... I'll continue to check it periodically over next hour and if it consistently works I'll file task.
I have a client that's really after me to implement "auto-save" in editor. I just noticed the autosave feature in TinyMCE, although that's not what most folks would call "autosave". ;-)
With NatEdit, I believe I found a very simple way to implement autosave.
maybe it's crude but seems to work.
I can trigger a click on "Save and Continue" button and it works fine in either WYSIWYG or regular NatEditor.
Now it's just a matter or setting timer to repeat.
[17:12]
LavrAutosave is both a blessing (when your browser crashes) and a curse (when you work on an important application which breaks if the topic is saved half edited).
So it is important to be able to disable it when you need to.
[17:15]
LynnwoodI personally would not like it.
Lavr - totally agree.
[17:16]
Lavrwhen I see people making meeting minutes online without saving I can see the autosave as a nice thing to have [17:16]
gac410y. I would be very reluctant to autosave. Crafting template changes, or wikiapps (ApacheConfigGenerator comes to mind) could be disasterous [17:16]
LavrExactly [17:17]
gac410Days??? strikeone keys will expire too. [17:17]
LavrSessions also [17:17]
Lynnwoody i know... you would think they could just save their topic... [17:17]
gac410 - the change to ControlWikiWordPlugin seems to be working beautifully. Well done!
impressive that you could identify possible issue in code so quickly.
[17:23]
LavrIt is quiet on my wiki. It is TIME!
Time for the big change. I am tgz-ing the data now.
[17:25]
LynnwoodAny suggestion about headline for task? "ControlWikiWordPlugin works inconsistently" is pretty vague. [17:26]
gac410Lynnwood: inconsistently with persistent perl? [17:26]
Lynnwoodok better... [17:26]
gac410Anytime inconsistent on fcgi, etc. almost has to be a global variable. And there was only one significant one $disabled. Rather quick leap. :)
gac410 does have a life, believe it or not, though sometimes he wonders. Gotta run get stuff done. back later. Good luck Lavr
[17:27]
LavrMy production site has a alert message now and save, edit and rest scripts are gone. We are in read only mode. Tgz being produced of all the data
Tomorrow a few will love me and a lot will hate me :-)
The faith of an admin
or is it Fate of an admin ? :-)
[17:31]
......... (idle for 40mn)
***vrurg has left [18:13]
LavrSo I have now a Foswiki 2.1 in production. Now waiting for the phones to ring. [18:23]
............. (idle for 1h0mn)
So far no angry customers [19:23]
....... (idle for 30mn)
gac410cool ... great news Lavr. [19:53]
........................ (idle for 1h55mn)
***KryoStoffer has quit IRC (*.net *.split)
alexlist has quit IRC (*.net *.split)
jayen_ has quit IRC (*.net *.split)
[21:48]

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