#foswiki 2017-01-30,Mon

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

WhoWhatWhen
***ChanServ sets mode: +o Lynnwood__ [01:31]
ChanServ sets mode: +o Lynnwood [01:39]
ChanServ sets mode: +o Lynnwood [01:46]
........ (idle for 39mn)
ChanServ sets mode: +o Lynnwood__ [02:25]
......... (idle for 41mn)
ChanServ sets mode: +o Lynnwood [03:06]
.................... (idle for 1h38mn)
gac410 has left [04:44]
.......................... (idle for 2h9mn)
ChanServ sets mode: +o cdot [06:53]
.............................. (idle for 2h26mn)
ChanServ sets mode: +o MichaelDaum [09:19]
GithubBot[distro] MichaelDaum pushed 1 new commit to Release02x01: https://git.io/vDTVI
distro/Release02x01 6c8cda0 MichaelDaum: Item14312: don't destroy dialogs on close
[09:27]
***GithubBot has left [09:27]
FoswikiBothttps://foswiki.org/Tasks/Item14312 [ Item14312: don't destroy dialogs on close ] [09:27]
GithubBot[distro] MichaelDaum pushed 1 new commit to master: https://git.io/vDTVY
distro/master 14426f0 MichaelDaum: Merge remote-tracking branch 'origin/Release02x01'
[09:27]
***GithubBot has left [09:27]
GithubBot[distro] MichaelDaum pushed 1 new commit to Item14288: https://git.io/vDTVa
distro/Item14288 ecf73a9 MichaelDaum: Merge branch 'master' into Item14288
[09:30]
***GithubBot has left [09:30]
FoswikiBothttps://foswiki.org/Tasks/Item14288 [ Item14288: rewrite to support pluggable edit engines ] [09:30]
GithubBot[distro] MichaelDaum pushed 1 new commit to Item14288: https://git.io/vDToU
distro/Item14288 5170d22 MichaelDaum: Item14288: resolve deferred obj on close
[09:42]
***GithubBot has left [09:42]
...... (idle for 25mn)
foswiki_irc7Hi
I'm having trouble with ClassificationPlugin.
I don't know if I have it set up correctly. Is there an on-line example that I could look at?
[10:07]
MichaelDaumfoswiki_irc7, have a look at https://demo.michaeldaumconsulting.com [10:08]
foswiki_irc7Will do - thank you [10:08]
MichaelDaumyw [10:11]
........ (idle for 38mn)
foswiki_irc7ClassificationPlugin::Core.pm line 107 has this: my $web = Foswiki::Plugins::DBCachePlugin::getCore->currentWeb() || $session->{webName}; [10:49]
FoswikiBothttps://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Plugins::DBCachePlugin [10:49]
foswiki_irc7and it is generating an error: Can't locate object method "currentWeb" via package "Foswiki::Plugins::DBCachePlugin::getCore" (perhaps you forgot to load "Foswiki::Plugins::DBCachePlugin::getCore"?) at ...
Should that be getCore()->currentWeb ?
FWIW, I'm running perl 5.12.3
[10:50]
MichaelDaumfoswiki_irc7, please upgrade DBCachePlugin/Contrib and ClassificationPlugin to the latest releases [10:52]
foswiki_irc7Will do. [10:52]
I already have latest ClassificationPlugin (6.01). I have DBCachePlugin 9.20, the latest is 10.00, ClassificationPlugin wants >= 4.10
Installing new DBCachePlugin&Contrib now
[10:59]
....... (idle for 31mn)
Thank you MichaelDaum, that fixed it :-) [11:30]
.................... (idle for 1h37mn)
***ChanServ sets mode: +o gac410 [13:07]
..... (idle for 23mn)
gac410MichaelDaum: ... I'm seeing intermittent issues with blog.foswiki.org
Failure is: 08:29:27.747 ReferenceError: jQuery is not defined 1 Foswiki213Beta2ReadyForTesting:313:1
<anonymous> https://blog.foswiki.org/Blog/Foswiki213Beta2ReadyForTesting:313:1
Happened yesterday after first browse to the site. Shortly again all was good. Now this am, failing again. Login dropdown is "inactive" and disqis is missing.
Could the pageoptimizer not be bundling everything needed?
[13:30]
MichaelDaumgac410, checking .... sec
I thinkg DisqusPlugin has got a missing constraint on jquery when adding its js to the page
[13:36]
gac410Yesterday when it happened, the page was "unstyled" like the css was missing, but a little later it came back. but refresh/ clear cache, etc. had no impact
Today it's styled, but the nat js pieces are missing "like" actions, login,
[13:38]
MichaelDaumhm no the addToZone constraints are just fine
could you create a task item and collect the screenshots of what it looks like, please?
[13:39]
gac410Okay
Damn... Discus is back now. This is really strange and the jquery error has cleared.
just like yesterday, it suddenly just cleared after some innocuous action
[13:39]
MichaelDaumas soon as the page is broken, collect a screenie, whether you are logged in or not, any js error, and any 404 file not found errors [13:44]
gac410okay will do. Have you observed the issue, or is it just me so far? [13:44]
MichaelDaumone possible scenario could be: the PageOptimizer cache is cleared ... but the page cache not
yes. under these circumstances.
css and js get combined and are stored in .../pub/System/PageOptimizerPlugin/cache.
then a cronjob clears the cache, however leaves the page cache intact
the css and js need to be re-combined to build up the PageOptimizer cache again. however this does not happen as the page cache just serves the page.
[13:44]
gac410Ah. I did try a refresh=cache on the page. I didn't *think* it worked, but maybe it did and I just wasn't observant enough. [13:46]
MichaelDaumnot sure whether I am clear. [13:46]
gac410No it makes sense. [13:46]
MichaelDaumuebera||, can you take a look at the cronjobs please: we need a ./view refresh=all right after the ./rest /PageCachePlugin/purgeCache
or maybe we just leave out the latter
[13:47]
gac410Do you still need a task? [13:48]
MichaelDaumas a reminder to do something about it, yes, that would be nice. thanks. [13:48]
gac410Okay. Against WebSite, or PageOptimizer or ? [13:49]
MichaelDaumI simply fixed&forgot it on another site ... ;|
PageOpti + PageCache
[13:49]
gac410okay [13:49]
uebera||MichaelDaum: The above call to view outputs the (main) page, is the output supposed to be redirected or discarded? [13:51]
MichaelDaumdev-null it :) [13:52]
GithubBot[GridLayoutPlugin] MichaelDaum pushed 1 new commit to master: https://git.io/vDkfF
GridLayoutPlugin/master 856cb3b MichaelDaum: Item14313: fixed css selector for responsive grid
[13:53]
***GithubBot has left [13:53]
FoswikiBothttps://foswiki.org/Tasks/Item14313 [ Item14313: grid fails to collapse ] [13:53]
uebera||done [13:53]
gac410Damn, now that I'm logged in (on Tasks web) I went back to blog to get the spelling of disquius for the task, and the bottom of page is gone again.
And refresh=cache fixes it.
so yes, it is indeed related to caching.
[13:54]
uebera||But ./view refresh=all has already been executed four minutes ago. [13:54]
gac410We may have crossed paths there.
The page *did* show the cached info on the bottom before I reloaded it.
But I didn't grab the timestamp. :(
[13:55]
MichaelDaummeanwhile foswiki.org 500'ed at me uploading a new GridLayoutPlugin [13:57]
gac410MichaelDaum: Will the fix be committed to PageCache, or PageOptimizer plugin. If both, I need two tasks to keep tracker happy :( [13:57]
MichaelDaumdon't we support "Component" to have a list of entries in it?
if in doubt PageOptimizerPlugin is the one to blame
[13:58]
gac410yes, But it falls into the "who closes the task" issue. When I release I close all tasks in the release notes.
If you release POP and close the task then my release notes are f'd up
[13:59]
MichaelDaumthe point is: POP has to clear the page cache when it receives a purge-cache request
at least thats what my thinking is so far
[13:59]
gac410I'll list both componenets, but will make a note to please not mix core & non-core commits on a single task. [14:00]
MichaelDaumactually, leave out PageCache from Component
it really is a POP issue
[14:00]
gac410Okay. That works for me :D
Hm Would putting the PageOptimizer piece into a dirty area make sense?
[14:01]
MichaelDaumno
it prevents the page from being cached in gzip'ed format
no dirty areas -> cache a gzip'ed html page
dirty area -> cache html in cleartext -> gzip on delivery
... after having recomputed the dirty areas
[14:03]
gac410interesting. Is the gzip process that slow? since apache can do it anyway? [14:05]
MichaelDaumyou basically want to prevent the cpu to do any work twice ... this includes gzipping [14:05]
gac410Item14314 [14:06]
MichaelDaumhaving apache gzipping text/plain means ... it does so again and again delivering the same bits and bytes [14:06]
FoswikiBothttps://foswiki.org/Tasks/Item14314 [ Item14314: Intermittent rendering issues on blog.foswiki.org ] [14:06]
MichaelDaumbetter is to cache the efforts and just send over the gzip'ed blob [14:06]
gac410Need to think about that then. Don't we enable mod_gzip, or is it mod_deflate automatically? [14:06]
MichaelDaumit kind of is bad practice for high end sites
sounds counterintuitive but it'
s true
we have js.gz and css.gz on disc already. these are served right away. so no gzip is required again.
^is NOT required again
ach ... first type then think
not my day
[14:07]
gac410I think we need to review our ApacheConfigGenerator ... we generate cfg that checks for existence of the .gz files and will rewrite the request to the gz version. [14:09]
MichaelDaumy [14:09]
gac410# <FilesMatch "\.(js|css)$"> RewriteEngine on RewriteCond %{HTTP:Accept-encoding} gzip RewriteCond %{REQUEST_FILENAME}.gz -f RewriteRule ^(.*)$ %{REQUEST_URI}.gz [L,QSA] [14:10]
MichaelDaumthis should be enabled by default ... not commented out [14:11]
gac410It's all commented out by default. On a production site I upgraded to 2.1.2, it was all coming out unstyled until I re-commented all those rules. [14:11]
MichaelDaumgawd, nginx does so by default without any extra provisions what so ever.
if it wasnt working then theres something else wrong in ACG
[14:12]
gac410That's scary. If you want to edit the js, then it is just magically ignoring your files? Or does it check file mtimes [14:13]
MichaelDaumediting js or css requires these files to be minified and recompressed.
you need some nodejs tools for that
[14:14]
***ChanServ sets mode: +o Lynnwood [14:14]
gac410yikes. That makes using nginx as a developer really difficult. I'm glad I'm on apache here [14:15]
MichaelDaumthis has got nothing to do with the web server. [14:15]
gac410You just said "gawd, nginx does so by default without any extra provisions what so ever." [14:15]
MichaelDaumy
do you agree minified and compressed css and js are better for web performance?
[14:15]
gac410Yes of course [14:16]
MichaelDaumthen you need to do so [14:16]
gac410I think we are mis-communicating. [14:16]
MichaelDaum:D [14:16]
gac410I'm saying that as I'm sitting here on my laptop editing js/css on my laptop for my laptop test apache server. I don't want to have to compress/minimize every time I try something.
And if I read "nginx does so by default" ... then that says if I were using nginx, I have to compress/minimize for every single edit/test cycle
Or am I missing somethig
ah. except pseudo-install symlinks the compressed to the uncompressed versions.
gac410 needs coffee and breakfast
[14:17]
MichaelDaumnginx (as well as apache with some extra rulez) will serve foobar.css.gz if it was found sitting next to a foobar.css [14:20]
gac410that's fine. And for a prod site, thats really good. [14:20]
MichaelDaumput the other way around: if there is no foobar.css.gz, then the webserver will send over the non-compressed foobar.css
all of our extension css and js comes in three forms: foobar.umcompressed.css, foobar.css and foobar.css.gz
[14:21]
gac410Okay, so as a dev, I just delete the .gz, or symlink it to the uncompressed, and I'll get the uncompressed for testing [14:21]
MichaelDaumthe first one is the human readable, the other two are used to be served by the webserver.
thats right
[14:21]
gac410Yes I do understand all this. [14:22]
MichaelDaumor you use the Makefile system ... like: edit -> save -> make [14:22]
gac410I'm making a tempest in the teapot here. [14:22]
MichaelDaumnote you need uglifyjs and cssmin [14:22]
gac410y I have all them ... I do build the releases :D
Which reminds me. Since they (uglify / cssmin) are installed outside of my apt-update ... I need to figure out how to update them before building a new release.
[14:22]
MichaelDaumat one point or the other I'd like to move away from Makefiles and use some real build tools straight outta npmjs.org [14:23]
gac410I never have ever touched make. [14:24]
MichaelDaumnpm update -g [14:24]
gac410I use BuildContrib build.pl release [14:24]
MichaelDaumI once was a C programmer and have written tons of makefile systems for projects ... roots [14:24]
gac410gac410 hopes that buildcontrib in it's compress phase does everything I need to do. As nobody has ever specifiged using make in Foswiki:Development/BuildingARelease [14:25]
FoswikiBothttps://foswiki.org/Development/BuildingARelease [ BuildingARelease ] [14:25]
gac410gac410 follows that topic pretty much step-by-step for every release. [14:26]
MichaelDaumit is there to support devs, not building a release.
when I am in pub/System/SomePlugin and edit some js, I just type "make" to update this directory's css and js.
[14:26]
gac410Other than you mentioning it on occasion, I don't ever remember seeing our documentation saying to run make. Not just BuildingARelease. [14:27]
MichaelDaumor "make jslint" to check code quality
yea and it probably should be replaced with something modern
[14:27]
gac410Our docs, especially the "contributed" stuff in support and development, really need to be thoroughly scrubbed.
Some of the platform installation guides are really out of date
anyway... I gotta get breakfast biab
[14:29]
MichaelDaum: ... I forgot, did you have a chance to check over my email to announce 2.1.3 Beta 2 [14:44]
MichaelDaumisnt it the one on the blog? [14:46]
gac410mostly I think there was some wording changes in the intro, and thank-you at the end [14:47]
........ (idle for 36mn)
cdotMichaelDaum: I used the npm build tools for a while.... then moved back to make. Found the npm tools flaky and immature :-(
may have improved more recently, of course.
[15:23]
MichaelDaumwhich one did you use? [15:25]
cdotjake
I tried gulp as well, but couldn't get started with it - I was using an older version of node.js at the time, and I think that was why.
if I was going to try again now, I'd use npm and just live with the limitations.
[15:30]
..... (idle for 22mn)
***ChanServ sets mode: +o Lynnwood__ [15:55]
...... (idle for 26mn)
cdotgac410: re: Item13785 - re creating a session. If you want code to crib, WorkflowPlugin/tools/workflow remind and it's associated module lib/Foswiki/Plugins/WorkflowPlugin/Mither.pm does what you want - which is basically "$Foswiki::Plugins::SESSION = new Foswiki( $Foswiki::cfg{AdminUserLogin} );" [16:21]
FoswikiBothttps://foswiki.org/Tasks/Item13785 [ Item13785: ExtensionInstaller fails to replace files under some conditions. ] https://trunk.foswiki.org/System/PerlDoc?module=Foswiki::Plugins::SESSION [16:21]
gac410Y. Already did that in a test version i've got stashed. Seems to be fine. [16:21]
cdot:-) [16:22]
gac410I don't have to create a new one though, just create one if Plugins::SESSION is empty
I think this strangeness may be something in Store, related to fcgi. Both f.o (RcsLite) and Lynnwoods site (PlainFile) have encountered it,
And lynnwood actually caught the topic "reverting" after a refresh. Is there anything that store does if for some reason it thinks the METAINFO is wrong?
Lynnwood confirmed thoug that he was installing using the Web interface, which *does* have a session, and does checkin the changes into the store
[16:22]
cdotyes, remember the store needs modification dates on the cache (.txt) to be correct, otherwise it may try to check in a new version and update TOPICINFO [16:28]
gac410The symptoms though seem only to happen when fcgi is involved. So somehow store may have "remembered" the wrong version?
In Lynnwoods case he even encountered that the new content was in the history in PFS rev 1, but the old content was in the .txt file
out-of-order.
[16:29]
cdotcould it be a race condition?
cdot is guessing
[16:30]
gac410Maybe I really don't know ... pulled hair over this a year ago on f.o, and never figured it out.
Difficult because I could never recreate it locally. Which hints that maybe a race is involved But on f.o, i could not avoid creating the issue.
And then the page cache even complicated it further.
Is calling Meta::save() enough to invalidate the cache, or does it have to be a save transaction. In Lynnwoods case, the page cache is not enabled
[16:30]
To answer my own question. yes, save() which calls saveAs() calls fireDependency in a finally section, so the cache gets handled [16:44]
Okay... I
MichaelDaum: Should the blog be "abbreviating" it's stories? ie the "read more" link doesn't really show any more other than the disqus comments. What is the "teaser" section all about.
One other "little thing" that confused me. The "MyName" link you click to get to the login screen renders my pointer as a "I" text position, not a -> finger icon that would tell me to click it.
[16:51]
MichaelDaumin general the "teaser" section holds the first paragraph or so as a kind of introduction to the story. the frontpage will then render all teasers ... or the full blog posting if there is no teaser
in this case the "read more" link will get you to a page holding the complete posting plus comments
[17:05]
gac410hm I did set a teaser section on the beta announcement, but the whole thing is rendering
Y, what you describe is what I expected, but not what I'm seeing.
The teaser section ends just before the "For Users" heading
[17:06]
MichaelDaumIve set the ONLYTEASER flag on the frontpage now. [17:08]
gac410Okay. ah.. .it's a setting. The rest of our old articles don't have teaser sections. (or at least the couple I checked) We probably could add them to shrink the front page a bit. [17:09]
MichaelDaumsure thing [17:09]
gac410If I edit an old article, will it pop to the top of the front page?
(Which I do Not want )
[17:09]
MichaelDaumno [17:10]
gac410okay good [17:10]
MichaelDaumthis is sorted by publication date, not modification or creation date [17:10]
gac410okay great. Thanks. Just making sure not to add to the may holes in my feet [17:11]
Okay, I've set the teaser sections on the visible blog stories. feel free to adjust :D My judgment on the marketing stuff is not the best. [17:19]
................... (idle for 1h32mn)
***ChanServ sets mode: +o Lynnwood [18:51]

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