Time Nick Message
04:24 * Francesca waves
07:40 marcelr hi #koha
07:45 reiveune hello
07:54 alex_a bonjour
08:10 gaetan_B hello
11:36 oleonard Hi #koha
11:38 oleonard I'm not sure what to do today since we fixed all the bugs yesterday.
12:44 Joubu @RM could you please push the follow-up on bug 16044?
12:44 huginn Joubu: I suck
13:31 oleonard "PERL6: Indentation: Indentation must be 4 spaces (no tab, no 2 or 6 or 8 spaces)"
13:32 oleonard We should decide whether we are going to stick to that. It doesn't belong in the guidelines if it's not going to be enforced.
14:42 oleonard Everyone is hung over from GBSD?
14:43 oleonard I must have missed the afterparty.
14:46 jcamins oleonard: what about 3-space indents? I think everything should be switched to 3-space indentation.
14:46 oleonard You might as well jcamins, it's a free-for-all out there.
14:47 oleonard Express your desire for 3-space indents by submitting lots of patches using 3 spaces. That's how we vote around here.
14:47 jcamins Disadvantage of this plan: need to write Perl.
14:48 * oleonard barricades the door to prevent cait from leaving again
15:05 tcohen hi
15:07 bag hi tcohen
15:13 tcohen Joubu: how can u have such variance?
15:14 Joubu between what?
15:14 Joubu 20 vs 100?
15:14 tcohen between the 10 interations on each test
15:15 Joubu tcohen: I don't know. It's a lxc container on my laptop. Not a proper server
15:15 tcohen it is probably related to caching :-D
15:16 Joubu tcohen: How that?
15:16 tcohen consecutive runs should should more cache usage
15:17 tcohen (less computing time)
15:17 tcohen but it depends on the worker you hit I guess
15:17 tcohen you should try running a long lasting single worker/process
15:17 tcohen to have more deterministic results
15:18 tcohen opt="$opt --server Starman -M FindBin --max-requests 50 --workers 4"
15:18 Joubu Do you think different workers are used?
15:19 tcohen between runs? i'd expect so
15:19 * cait waves
15:25 Joubu tcohen: heh,
15:25 Joubu yes, good catch
15:25 Joubu indeed it's much more close now
15:25 Joubu so I need to redo everything... :-/
15:28 Joubu tcohen: ha no, I get 7.5 and 8.4
15:29 Joubu 9.469+8.155+8.185+8.407+7.560+7.811+7.657+7.795+7.631+7.125
15:29 wahanui 79.795
15:29 oleonard wahanui++
15:33 gaetan_B bye
15:34 pianohacker Joubu: what benchmarks are you talking about?
15:35 pianohacker Joubu: also, as a minor side note, could I ask you to use my bywater email for bugzilla, please? (jweaver@)
15:35 Joubu pianohacker: see bug 16140
15:35 huginn 04Bug http://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=16140 enhancement, P2, ---, jweaver, Needs Signoff , Only clear L1 cache when needed
15:35 Joubu pianohacker: yep!
15:36 pianohacker sweet!
15:37 pianohacker Though the first set of benchmarks makes much more sense than the first, to me at least
15:37 Joubu pianohacker: the first or the first? :)
15:38 pianohacker wow, I should not try to talk code before espresso
15:38 pianohacker first makes more sense than the second, but I see your note about 20 vs 100 (search results?)
15:38 Joubu yes, the pref numsearchresults
15:39 oleonard pianohacker: Maybe do a bulk update of bugs assigned to 'pianohacker'? That might be what triggers the autocomplete suggestion?
15:39 pianohacker definitely looks like we should push both, especially since 20 is the default
15:39 pianohacker oleonard: that's worth a shot, thanks
15:39 pianohacker Joubu: I'll try to duplicate your benchmarks sometime today
15:40 Joubu pianohacker: did you notice what tcohen said? use only 1 worker
15:41 pianohacker yeah, that makes a big difference in consistency
15:41 Joubu I don't understand why the L1 cache does not bring perf gain directly
15:42 Joubu (16044)
15:42 pianohacker this plack or no?
15:42 Joubu yes of course, plack + memcache
15:43 pianohacker I _think_ that my second patch on 16410 honestly has a bigger impact than the first for the specific case of search, because that code hits the MARC frameworks so hard and deep cloning is really expensive for them
15:46 pianohacker your comment makes me happy for how far things have come :)
15:47 Joubu :)
15:48 pianohacker Joubu: one thing that might be interesting would be to test 16140 + 16088 but with the first patch ("Bug 16140: Only clear L1 cache when needed") reverted... it may be that the intelligent L1 flushing is having less of an effect than removing the deep cloning
15:48 huginn 04Bug http://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=16140 enhancement, P2, ---, jweaver, Needs Signoff , Only clear L1 cache when needed
15:50 Joubu ha yes, indeed. Good idea
15:50 oleonard Anyone want to apply to be my boss? https://www.myacpl.org/jobs/
15:51 pianohacker if you're benchmarked out for the day, I'll be repeating your benchmarks with your test file and will throw that in the mix
15:51 Joubu pianohacker: or better, compare 16088 with first patch of 16140
15:52 pianohacker Joubu: do you think 16140 should be split again?
15:52 pianohacker it really is two separate things
15:53 Joubu Yes it is, but I am a bit afraid to split everything up and don't manage to stick back together :p
15:54 pianohacker fair :)
15:54 Joubu pianohacker: as I said yesterday, 1 bug report per idea with test coverage is the best approach
15:57 Joubu pianohacker: your first patch is terrible for perfs :)
16:01 pianohacker worried that would be the case... I'll run a big long suite of benchmarks later today
16:19 reiveune bye
17:12 Joubu pianohacker: it would be awesome to get the selenium tests pushed, we could add more!
17:15 pianohacker Yes indeed :)
17:16 Joubu we could catch easily if a feature/patch introduce a gap in the perfs
17:30 Joubu have a good week-end #koha
17:44 kidclamp you too Joubu
19:29 tcohen hi
19:34 tcohen Joubu: around?
20:13 misilot Is there part of the ILS-DI API the ability to do a live inventory? ie: I can make a webform that submits the barcode and it returns the status (checked out / missing / lost ) and updates the last seen date and checks it in if checked out / marks as not missing?
20:43 tcohen hi
20:45 bag hi
21:16 huginn New commit(s) kohagit: Bug 15838 - Subscription duplicating: Reset fields from SubscriptionDuplicateDroppedI... <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=873a49f13b79bf1f5f7163f217cfc3a317ce602f> / Bug 16145: Regression: Bug 15632 broke display of library name on circulation messages <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=a186dcfd84da4b22fb59eb97d9889f9d7fa3571b> / Bug 16084: [QA Follow-up] Correct
21:36 huginn New commit(s) kohagit: Bug 16142: Making 'Order' a button for New Order Suggestions <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=9712a7448e94df8e22debca0670fa4e604ea0d4f> / Bug 14076: Do not use CGI->param in list context - opac-authorities-home.pl <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=3fa2b10150a9ea2db2897be1246cba3785c55e55>
21:56 huginn New commit(s) kohagit: Bug 15413: Adding colons where appropriate <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=e428ff001d7e877b4f666444add9236fcb491d38> / Bug 16139: Renaming 'Unseen since' column to 'Last seen' <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=56cb727aecf9b55e95be6f4dc5218b472e4dcca3> / Bug 16141: Making Transfers to Receive action a button <http://git.koha-community.org/gitweb/?p=koha.git;a=co
22:06 huginn New commit(s) kohagit: Bug 14304: RDA: OPAC display link in XSLT for 264 field to reflect Zebra indexing <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=0303fa7de6fd8f9552440fa9824e402daf8e24b8> / Bug 14304: RDA: Intranet display link in XSLT for 264 field to reflect Zebra indexing <http://git.koha-community.org/gitweb/?p=koha.git;a=commitdiff;h=2f0fc5f45abdd9ee31a2bebf40a634a397fdfd93>
22:30 Francesca_ @wunder christchurch, new zealand
22:30 huginn Francesca_: The current temperature in Christchurch, New Zealand is 14.0°C (11:00 AM NZDT on March 26, 2016). Conditions: Mostly Cloudy. Humidity: 82%. Dew Point: 11.0°C. Pressure: 30.06 in 1018 hPa (Steady).