06:03 <PaulFertser> borneoa_: good points. The mails on new patches are sent by a hook to the mailing list, so that's special and indeed I see in the logs they fail, will debug.

06:04 <PaulFertser> borneoa_: disabled the commit message checks altogether now.

06:42 <borneoa_> PaulFertser: thanks. I think the length of the commit message will be checked by checkpatch during the build, so can be skipped at push.

06:49 <PaulFertser> borneoa_: thought exactly the same and disabled the gerrit plugin.

06:50 <PaulFertser> borneoa_: will fix the mailing list notification but for that some change needs to be at least updated (new changeset version created) as debug logs for the hooks were off.

06:51 <borneoa_> PaulFertser: I can update one of my patches, but not before 1 hour

06:55 <PaulFertser> borneoa_: np, I'll be commuting for at least an hour soon anyway.

07:00 <PaulFertser> borneoa_: actually, nevermind, it's just bashism in /bin/sh patch.

07:02 <Fleck> morning!

07:06 <PaulFertser> borneoa_: so no worries: reproduced, understood, will be fixed soon, thanks for the report :)

07:06 <PaulFertser> borneoa_: old machine had bash for /bin/sh

07:26 <PaulFertser> borneoa_: patch should be fixed

07:27 <borneoa_> PaulFertser: thanks!

07:28 <PaulFertser> borneoa_: only testing will show if it's fully fixed, so will be waiting for the next new patch to be pushed.

08:29 <borneoa_> PaulFertser: just sent a new patch. The warning of 50 is over.

08:30 <borneoa_> PaulFertser: but still no email!

08:37 <PaulFertser> borneoa_: back to the terminal, will check it soon

08:46 <PaulFertser> borneoa_: yes, enabling logs showed what changed in hooks syntax, fixing soon.

09:23 <PaulFertser> borneoa_: fixed the script, manually rerun it, seems the mail is ok now.

09:24 <borneoa_> PaulFertser: yes, I have received it

09:25 <PaulFertser> Twice I guess, I seem to have run it one extra time.

09:25 <borneoa_> PaulFertser: thanks. No, only once

09:25 <borneoa_> PaulFertser: yes, arrived the second

09:27 <PaulFertser> borneoa_: do you have any strong opinions on migrating the static blog from sf.net to self-hosted on same machine?

09:28 <PaulFertser> I think current jenkins config is not triggering builds on push to master, I will investigate later.

09:29 <borneoa_> PaulFertser: no, sf used to be slow but now seems ok. Maybe using static blob helped

13:10 <PaulFertser> Good trend, warnings were reduced by one :) https://build.openocd.org/job/openocd-clang/

13:12 <PaulFertser> https://build.openocd.org/job/openocd-clang/lastBuild/clang/fixed/

15:22 <borneoa_> PaulFertser: I have noticed that at least on Firefox (not check on other browsers), when I switch to another tab, then after a while return on the tab with the new gerrit page, the tab is fetched again! Odd! Analysing the traffic, got that all the fonts are sent by garrit with header "Expires:" with actual time. It means the whole page get immediately invalidated from browser cache after it's loaded. Anything that can be done to avoid it?

15:23 <Haohmaru> o_O

15:29 <borneoa_> PaulFertser: same for css files, JavaScript, favicon.

15:31 <Haohmaru> this gerrit thing is sorta like gitlab/github/sourceforge/etc.. right?

15:34 <Haohmaru> seems yes..

15:41 <borneoa_> Haohmaru: yes. It's where we review the patches for openocd code.

15:42 <Haohmaru> so it's doing weird things, naughty HTML, but you're not in control of it?

16:08 <borneoa_> Haohmaru: the server was upgrade to more recent version during the weekend. Should be some wrong setting in the http server. No, I don't control it.

16:08 <Haohmaru> contact $webadmin then? ;P~

16:31 <PaulFertser> borneoa_: will take a look, thanks for detailed report, very useful.

19:27 <PaulFertser> borneoa_: "Note: If there is a Cache-Control header with the max-age or s-maxage directive in the response, the Expires header is ignored. " https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Expires

19:33 <PaulFertser> Cache-Control: private, max-age=900, must-revalidate

19:34 <borneoa_> PaulFertser: max age is set to 900 seconds, so 15 minutes. But I see the reload after less than that when switch tab. Investigating...

19:35 <PaulFertser> Probably gertty is going to work much better than the web interface :)

19:37 <borneoa_> PaulFertser: it's the JSON data that are reloaded

19:37 <borneoa_> PaulFertser: and it triggers the whole redraw of the page

19:37 <PaulFertser> borneoa_: probably your web browser terminates the ajax connection when the tab is in the background?

19:46 <borneoa_> PaulFertser: I'm comparing with another gerrit that don't have this behavior. I'm changing tab from openocd to this other back and forth. Only openocd reloads. Now I'm comparing the http headers

19:47 <PaulFertser> borneoa_: and it's trivial to patch gertty to show conflicts, e.g. I see plenty of conflicts just viewing 6446

19:48 <PaulFertser> (after remiving is:mergeable from its conflicts query)

19:49 <borneoa_> PaulFertser: openocd reply with "Connection: keep-alive" and "Keep-Alive: timeout=5, max=100". The other has "Connection: close". Not sure this is the reason

21:47 <borneoa_> PaulFertser: so, it's a feature! Undesired by me, but it's a feature. RELOAD_DASHBOARD_INTERVAL_MS=10*1000;

21:48 <borneoa_> PaulFertser: only the dashboard is reloaded if you change tab and return after more than 10 seconds.

22:01 <borneoa_> PaulFertser: it was introduced as experimental feature in gerrit 3.3.2 and become the default in 3.3.3. Really ugly feature.