|
2018-10-17 01:58:22
|
<revi>
|
lol
|
|
2018-10-17 01:58:32
|
<revi>
|
such bans
|
|
2018-10-17 02:00:21
|
<AlexZ>
|
indeed
|
|
2018-10-17 14:00:49
|
<wm-bot>
|
Technical Advice IRC meeting starting in 60 minutes in channel #wikimedia-tech, hosts: @Thiemo_WMDE & @chiborg - all questions welcome, more infos: https://www.mediawiki.org/wiki/Technical_Advice_IRC_Meeting
|
|
2018-10-17 14:50:15
|
<wm-bot>
|
Technical Advice IRC meeting starting in 10 minutes in channel #wikimedia-tech, hosts: @Thiemo_WMDE & @chiborg - all questions welcome, more infos: https://www.mediawiki.org/wiki/Technical_Advice_IRC_Meeting
|
|
2018-10-17 14:53:35
|
<mainframe98>
|
Hi, I'm sorry for breaking in so abrupt, but I just noticed that the messages in the sidebar on MediaWiki.org no longer show the right text, but the message name. Has a task for that already been filed?
|
|
2018-10-17 14:53:55
|
<mainframe98>
|
They now show the exact text as defined on [[mw:MediaWiki:Sidebar]]
|
|
2018-10-17 14:54:42
|
<Krenair>
|
woah
|
|
2018-10-17 14:54:47
|
<Krenair>
|
that's no good
|
|
2018-10-17 14:55:20
|
<Reedy>
|
sweet
|
|
2018-10-17 14:56:10
|
<Krenair>
|
mainframe98, that doesn't appear at https://phabricator.wikimedia.org/T191072
|
|
2018-10-17 14:56:16
|
<Reedy>
|
How long has that been broken for?
|
|
2018-10-17 14:56:39
|
<mainframe98>
|
Reedy: No longer than today, I didn't see it yesterday.
|
|
2018-10-17 14:57:24
|
<mainframe98>
|
Krenair: That's where I checked, but I wanted to make sure I didn't miss anything
|
|
2018-10-17 14:57:35
|
<Krenair>
|
that version has been there since some time yesterday
|
|
2018-10-17 14:58:21
|
<Krenair>
|
mainframe98, want to make a ticket?
|
|
2018-10-17 14:58:25
|
<mainframe98>
|
I'll create a task then, and add it as a deployment blocker. I'll quickly check against MW-vagrant as well, I have than one running right now
|
|
2018-10-17 14:58:25
|
<Krenair>
|
or shall I?
|
|
2018-10-17 14:58:31
|
<Krenair>
|
yeah
|
|
2018-10-17 14:59:09
|
<Krenair>
|
I can't find anything new in search about the sidebar either
|
|
2018-10-17 15:00:30
|
<chiborg>
|
Welcome to the Technical Advice IRC Meeting, Thiemo will join shortly.
|
|
2018-10-17 15:00:32
|
<mainframe98>
|
Vagrant now spits a bunch of notices about MediaWiki:Sidebar: Cannot create a user with no name, no ID, and no actor ID [Called from MediaWiki\Revision\RevisionStore::newRevisionFromRow in /vagrant/mediawiki/includes/Revision/RevisionStore.php at line 1784] in /vagrant/mediawiki/includes/debug/MWDebug.php on line 309, so I suspect it is related
|
|
2018-10-17 15:01:43
|
<Thiemo_WMDE>
|
I'm here for the Technical Advice IRC meeting together with chiborg (Gabriel). We both work for WMDE fulltime and are here for 60 minutes to answer any questions.
|
|
2018-10-17 15:02:14
|
<Thiemo_WMDE>
|
MGC prepared one questions on the wiki page: https://www.mediawiki.org/wiki/Technical_Advice_IRC_Meeting
|
|
2018-10-17 15:02:24
|
<Krenair>
|
mainframe98, write that on the ticket too :)
|
|
2018-10-17 15:02:34
|
<mainframe98>
|
Alright, task created: T207288, I'll edit the stacktrace to look better in a few moments
|
|
2018-10-17 15:02:35
|
<stashbot>
|
T207288: Text in the Sidebar does no longer show the message text, only the message name - https://phabricator.wikimedia.org/T207288
|
|
2018-10-17 15:03:30
|
<Thiemo_WMDE>
|
Since discussing MGC's questions might take a while, are there possibly shorter things to discuss ones first?
|
|
2018-10-17 15:04:58
|
<mainframe98>
|
I have one. I have an extension that stores stuff in cache. It should be available everywhere. So I
|
|
2018-10-17 15:05:07
|
<bawolff>
|
mainframe98: your error sounds like its related to the actor migration...
|
|
2018-10-17 15:05:11
|
<bawolff>
|
maybe
|
|
2018-10-17 15:05:49
|
<mainframe98>
|
*'ve picked WANObject cache. Imagine a wiki that is not part of a family. Would the data still get cached, or do I have to pick a different cache depending if the wiki is part of a wiki family
|
|
2018-10-17 15:05:58
|
<mainframe98>
|
(and thus has WANObjectCaching configured)
|
|
2018-10-17 15:06:48
|
<MGC|mobil>
|
Technical Advice Meeting should be here and now, aren't they? Just to be sure
|
|
2018-10-17 15:07:25
|
<bawolff>
|
mainframe98: Generally speaking, caches are tied to the current wiki (via wfWikiId())
|
|
2018-10-17 15:07:38
|
<bawolff>
|
unless you use whatever the modern equivalent of wfGlobalCacheKey()
|
|
2018-10-17 15:07:52
|
<bawolff>
|
For example, rate limits are generally global, but most everything else is not
|
|
2018-10-17 15:07:58
|
<chiborg>
|
MCG|mobil yes, we'r here
|
|
2018-10-17 15:08:31
|
<mainframe98>
|
bawolff: The thing is I need to store stuff based on the wiki set, configuration for a specific set, which should be available to all wikis in the set, but they aren't necessarily in the same datacentre/cluster
|
|
2018-10-17 15:08:32
|
<bawolff>
|
mainframe98: The modern version is $myCacheObject->makeGlobalKey()
|
|
2018-10-17 15:09:03
|
<mainframe98>
|
bawolff: That's the one I use. However, using that on a LocalServerObject cache doesn't make the cached object available for other wikis
|
|
2018-10-17 15:09:25
|
<bawolff>
|
That's true, because LocalServerObject cache is for the current server only
|
|
2018-10-17 15:09:41
|
<Thiemo_WMDE>
|
mainframe98: Can't you regenerate the same data and cache them again on the other server? Why do you want to share the same cache?
|
|
2018-10-17 15:09:42
|
<bawolff>
|
It not only wouldn't be available for other wikis, it wouldn't be available for other servers serving the same wiki
|
|
2018-10-17 15:10:11
|
<MGC|mobil>
|
What can you do to get your patches reviewed? Getting Code Reviews is a major problem, sometimes for half an year or longer. This isn't fun.
|
|
2018-10-17 15:10:22
|
<bawolff>
|
mainframe98: For that use case, i think you are supposed to use MainObjectStash
|
|
2018-10-17 15:10:29
|
<bawolff>
|
but ideally you can just regenerate
|
|
2018-10-17 15:10:40
|
<Thiemo_WMDE>
|
MGC|mobil: Thats a long topic to talk about. Let's first try to understand mainframe98's problem.
|
|
2018-10-17 15:10:47
|
<MGC|mobil>
|
Okay
|
|
2018-10-17 15:11:09
|
<MGC|mobil>
|
I've got time
|
|
2018-10-17 15:11:10
|
<bawolff>
|
Also, typically, with multi-dc stuff (as imagined by MW), all datacenters serve the same set of wikis, they just have (partially) separte cache sets
|
|
2018-10-17 15:11:32
|
<bawolff>
|
MGC|mobil: Well, I've been waiting like 5 years for code review to be fun, hasn't happened yet ;)
|
|
2018-10-17 15:12:14
|
<Thiemo_WMDE>
|
I think mainframe98 does have a lot to look up now (MainObjectStash was mentioned above).
|
|
2018-10-17 15:12:38
|
<bawolff>
|
mainframe98: Are you actually using MW's multi-DC stuff (asking out of curiosity because i didn't think anyone outside of Wikimedia used it)
|
|
2018-10-17 15:12:46
|
<mainframe98>
|
So, based on that I would conclude I misunderstand caching in MW somewhat. In that case, I'll go with regenerating the objects and stashing them locally. I'll have a look at the MainObjectStash.
|
|
2018-10-17 15:13:11
|
<bawolff>
|
(by MainObjectStash, I mean MediaWikiServices::getInstance()->getMainObjectStash() )
|
|
2018-10-17 15:13:12
|
<mainframe98>
|
bawolff: Not personally, but if I ever want the extension to be WMF ready, I'd better be prepared
|
|
2018-10-17 15:13:35
|
<bawolff>
|
mainframe98: Caching is mediawiki is super complicated and there are like billions of layers
|
|
2018-10-17 15:13:48
|
<Thiemo_WMDE>
|
mainframe98: It's not like we have multiple thousand servers, and you are probably not caching gigabytes worth of information.
|
|
2018-10-17 15:13:48
|
<chiborg>
|
MGC|mobil Is it "all'" patches or for specific projects?
|
|
2018-10-17 15:14:43
|
<bawolff>
|
mainframe98: Unless you have specific requirements for persisting cache across multiple datacenters, I would just use the main cache, which is per DC. It will probably work better generally. Don't fall into the trap of premature optimization
|
|
2018-10-17 15:14:52
|
<mainframe98>
|
Thiemo_WMDE: Nah, it should be a few hundred kb in max.
|
|
2018-10-17 15:15:09
|
<Thiemo_WMDE>
|
mainframe98: Sharing a cache between servers sounds like something that only makes sense if the amount of cached data or instances is massive.
|
|
2018-10-17 15:16:00
|
<d3r1ck>
|
o/
|
|
2018-10-17 15:16:21
|
<chiborg>
|
d3r1ck o/
|
|
2018-10-17 15:16:32
|
<Thiemo_WMDE>
|
Ok, back to MGC|mobil: "How to get Code review for patches just noone reviews?"
|
|
2018-10-17 15:17:04
|
<mainframe98>
|
Thiemo_WMDE: It is used in a serious hot-code path, on every request, so I wanted to use the big guns. As bawolff says, premature optimisation is not good, so I'll use the MainObjectStash first, and see where I'll end up. Thank you all!
|
|
2018-10-17 15:18:18
|
<bawolff>
|
Usually sharing cache between datacenters is mostly used as a hack when you are really doing something that's not a cache, and should be actually using the db (At least in my experiance)
|
|
2018-10-17 15:19:51
|
<Thiemo_WMDE>
|
My approach to give an answer to MGC|mobile's question would be more like asking how he ended running into this situation in the first place? What was the motivation to upload a patch to Gerrit (I assume) without knowing if it would get review?
|
|
2018-10-17 15:20:00
|
<Thiemo_WMDE>
|
See, in an ideal world (I know we don't live in one, but please bare with me) you know who is going to review your code before you write it.
|
|
2018-10-17 15:21:14
|
<bawolff>
|
Thiemo_WMDE: That's pretty far from the world I live in
|
|
2018-10-17 15:21:20
|
<MGC|mobile>
|
Yeah
|
|
2018-10-17 15:21:24
|
<Thiemo_WMDE>
|
See, Gerrit is not a file system where you upload stuff and that's it. It's a tool used by people. In an ideal world, the first step is to get to know these people, and only then start giving them stuff to do.
|
|
2018-10-17 15:21:48
|
<bawolff>
|
Maybe something like 5% of my patches I know who is going to review them beforehand
|
|
2018-10-17 15:21:57
|
<mainframe98>
|
Thiemo_WMDE: That would imply the people are there, and available
|
|
2018-10-17 15:22:09
|
<MGC|mobile>
|
I just make code improvements, quite often to not actively maintained extensions.
|
|
2018-10-17 15:22:21
|
<bawolff>
|
And I actually work for WMF, where actually knowing who is going to review my patches is likely (compared to volunteers where that seems very unlikely)
|
|
2018-10-17 15:22:38
|
<Thiemo_WMDE>
|
From the perspective of people who are potentially willing to review patches (I for example do this a lot) a patch that comes out of nowhere (no Phabricator ticket exists, nobody talked to nobody first) puts workload on people who never asked for something like that.
|
|
2018-10-17 15:22:46
|
<MGC|mobile>
|
Or to extensions that I maintain, which is basically the same situation if you want to avoid self-review
|
|
2018-10-17 15:23:06
|
<mainframe98>
|
The thing is, with extensions, there are maintainers no longer maintaining, a few central people having a load of stuff on their plate, or others who simply can't properly review the code because they don't have the expertise
|
|
2018-10-17 15:23:24
|
<bawolff>
|
With random extensions, I think part of the problem is that mediawiki core devs don't really want to interfere with other people's random extensions, unless they know they are essentially abandoned
|
|
2018-10-17 15:23:58
|
<Thiemo_WMDE>
|
I don't believe this is about "expertise". If you need an "expert" to review code because a codebase is so complicated nobody else would understand it, you are doomed anyway.
|
|
2018-10-17 15:25:02
|
<bawolff>
|
While there is a min level of programming experience required
|
|
2018-10-17 15:25:21
|
<bawolff>
|
A surprising amount of non-programmers sometimes want to review patches they don't understand at all
|
|
2018-10-17 15:25:37
|
<Thiemo_WMDE>
|
Again, might I ask what the motivation is to upload patches to "random", "unmaintained" extensions? Honestly, what do you expect?
|
|
2018-10-17 15:26:16
|
<mainframe98>
|
Thiemo_WMDE: Helping 3rd party users who use the extension?
|
|
2018-10-17 15:26:43
|
<Thiemo_WMDE>
|
bawolff: Every review is helpful, review by non-experts even more than reviews by experts that just click +2 with no comment. If new people can't understand the code, it is *always* the code who is wrong, not the new people.
|
|
2018-10-17 15:27:07
|
<MGC|mobile>
|
Thiemo: I usually provide code to answer feature requests and bug reports. Before adjusting Core, I present my ideas in a task if this hasn't been done already and wait for feedback. However, it's not that rare that Phab tasks are stale for years anyway, who can tell if such features are wanted...
|
|
2018-10-17 15:27:23
|
<bawolff>
|
Thiemo_WMDE: I think you haven't seen how useless some reviews are
|
|
2018-10-17 15:27:50
|
<Thiemo_WMDE>
|
mainframe98: Ok, that a wonderful motivation. It essentially means you are then adopting this extension, and should get +2 rights. And yes, if you are the only active developer with +2 rights, please self-merge your patches.
|
|
2018-10-17 15:27:55
|
<bawolff>
|
Reviews of the form, "I really want feature X, so I'm saying I reviewed this, even though I didn't read the code and I don't know how to program"
|
|
2018-10-17 15:28:18
|
<Guest12345>
|
I have a Lua issue. Is it possible to make a template accepting other templates as arguments? Does it exist somewhere (for inspiration)?
|
|
2018-10-17 15:28:28
|
<Thiemo_WMDE>
|
bawolff: That really gross to say, both for me (I have seen everything) as well for the people investing their time to review other peoples code, trying to make it better.
|
|
2018-10-17 15:28:37
|
<MGC|mobile>
|
Thiemo: These are extensions where Wikiapiary tells 100s of Wikis use them, but WMF doesn't. These extensions need attention, but are left alone.
|
|
2018-10-17 15:29:42
|
<Thiemo_WMDE>
|
MGC|mobile: That's not a Wikimedia issue then, isn't it? Still the same ideas apply: Find at least a 2nd developer willing to maintain the extensions, get both +2 rights, and merge each others patches.
|
|
2018-10-17 15:30:10
|
<d3r1ck>
|
Guest12345: https://meta.wikimedia.org/wiki/Help:Advanced_templates
|
|
2018-10-17 15:30:28
|
<Thiemo_WMDE>
|
bawolff: I'm *very* sure nobody in the world ever wrote a review comment like this. This is you not respecting the reviewers time and effort.
|
|
2018-10-17 15:30:57
|
<MGC|mobile>
|
If you need 5 years to find the first one, from where should you get the second one?
|
|
2018-10-17 15:31:14
|
<Guest12345>
|
d3r1ck: Thank you, i'll read it
|
|
2018-10-17 15:31:23
|
<d3r1ck>
|
:+1:
|
|
2018-10-17 15:31:42
|
<bawolff>
|
I'm not suggesting that comment was written literally
|
|
2018-10-17 15:31:46
|
<Thiemo_WMDE>
|
You said the extension is used on hundreds of wikis. Somebody using it should be able to maintain it.
|
|
2018-10-17 15:32:28
|
<chiborg>
|
MGC|mobile So one answer to your question would be to ask the WMF if they have a clear stance on extensions that are not part of the WMF projects. I'm not sure about goals and missions, but I guess someone in WMF (probably a person in charge of volunteer efoort) might have an answer. And if that answer is "we don't support non-wmf-project extensions with code reviews", that's can be the start of
|
|
2018-10-17 15:32:28
|
<chiborg>
|
negotiating for +2 rights.
|
|
2018-10-17 15:33:03
|
<Thiemo_WMDE>
|
bawolff: Well, ok. So you get a "+1" with a comment that makes clear all the person wants is the feature, but probably didn't even looked at the code? Thats indeed not worth much, but also doesn't hurt, in my opinion.
|
|
2018-10-17 15:33:03
|
<MGC|mobile>
|
In the end, probably a mediawiki/* +2 person will merge it at some point.
|
|
2018-10-17 15:33:54
|
<MGC|mobile>
|
And this third party maintaining thing is happening quite rarely in my experience
|
|
2018-10-17 15:34:09
|
<bawolff>
|
dev relations will help new people find reviewers for their patches, but generally speaking WMF won't help with code review for third party extensions (for that matter, probably not much help for mediawiki core either)
|
|
2018-10-17 15:34:39
|
<Thiemo_WMDE>
|
I'm pretty sure there is no budget planned for extensions that are not deployed on WMF servers. Why should it? This is something the MediaWiki community should discuss, and MediaWiki != Wikimedia.
|
|
2018-10-17 15:35:10
|
<MGC|mobile>
|
chiborg: Currently, the situation is weird. Wikimedia people will often merge your code, but technically, they do it in their free time.
|
|
2018-10-17 15:35:36
|
<Thiemo_WMDE>
|
Some WMF and WMDE employees are investing time in such reviews anyway. But this is their personal decision.
|
|
2018-10-17 15:35:51
|
<Thiemo_WMDE>
|
Exactly.
|
|
2018-10-17 15:36:12
|
<bawolff>
|
Wikimedia people's free time makes the Wikimedia world go round
|
|
2018-10-17 15:36:33
|
<Thiemo_WMDE>
|
MGC|mobile: Do you have a specific example in mind, by the way? The Arrays or Variables extensions?
|
|
2018-10-17 15:36:38
|
<bawolff>
|
Even in mediawiki core, there's a shocking amount of important stuff that is technically "free time" work
|
|
2018-10-17 15:37:12
|
<chiborg>
|
Ah, now I understand. So the problem is more like "How to find community members, who are not necessarly part of WMF/WMDE, that have some technical expertise, for asking them to review my code"?
|
|
2018-10-17 15:37:13
|
<MGC|mobile>
|
Variables currently works out just fine, thanks to you
|
|
2018-10-17 15:37:58
|
<MGC|mobile>
|
Currently, Arrays is a problem for me.
|
|
2018-10-17 15:38:02
|
<MGC|mobile>
|
https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/Arrays/+/405057/
|
|
2018-10-17 15:38:03
|
<bawolff>
|
chiborg: prayer
|
|
2018-10-17 15:38:42
|
<MGC|mobile>
|
And Loops will probably be a problem in a few weeks.
|
|
2018-10-17 15:38:54
|
<chiborg>
|
Is there some data somewhere where you could search for community members in terms of - last contributed, first commit, popular repos. I guess something like that could be extracted from gerrit.
|
|
2018-10-17 15:39:27
|
<chiborg>
|
Would that help to get at least a rough list of potential people?
|
|
2018-10-17 15:39:39
|
<Thiemo_WMDE>
|
Another factor I find super-critical is how complicated patches are. See, when I review "random" patches my idea is to kick-off something, but I can't be the one who keeps it running. I need small, concise patches that take only a few minutes to review.
|
|
2018-10-17 15:40:34
|
<bawolff>
|
chiborg: https://www.mediawiki.org/wiki/Development_statistics might be helpful
|
|
2018-10-17 15:43:29
|
<Thiemo_WMDE>
|
MGC|mobile: https://gerrit.wikimedia.org/r/405057 can be made a lot smaller by splitting in into 2 patches, I believe. The first does nothing but extract the ExtArray class into it's own file. The 2nd patch changes the extension registration.
|
|
2018-10-17 15:43:52
|
<MGC|mobil_>
|
Network kicked me after the Development statistics
|
|
2018-10-17 15:44:41
|
<Thiemo_WMDE>
|
It's never a good sign to find a patch with 23 patch sets and +/- 1300 lines. Sane people immediately close the browser tab when they see this. :-(
|
|
2018-10-17 15:45:40
|
<MGC|mobil_>
|
You are right, but splitting doesn't really make it less complicated
|
|
2018-10-17 15:46:30
|
<MGC|mobil_>
|
In this case at least
|
|
2018-10-17 15:46:41
|
<MGC|mobil_>
|
Maybe psychology?
|
|
2018-10-17 15:47:44
|
<MGC|mobil_>
|
These are patches started from people with low experience, that's a problem, but they should be fine now
|
|
2018-10-17 15:50:20
|
<MGC|mobil_>
|
WiFi again ^^
|
|
2018-10-17 15:52:41
|
<MGC|mobil_>
|
Well, I'm aware we can't solve this problem here
|
|
2018-10-17 15:53:18
|
<MGC|mobil_>
|
I think a "Gerrit-Aklapper" might be helpful however
|
|
2018-10-17 15:55:47
|
<Thiemo_WMDE>
|
MGC|mobil_: The patch is +77 lines now, and I technically changed *nothing* it does.
|
|
2018-10-17 15:56:07
|
<Thiemo_WMDE>
|
https://gerrit.wikimedia.org/r/405057
|
|
2018-10-17 16:02:07
|
<MGC|mobil_>
|
Yeah, I should probably use this feature more
|
|
2018-10-17 16:03:25
|
<chiborg>
|
Thanks everyone for participating and for the lively discussion. I learned a lot! This concludes todays Technical Advice Meeting.
|
|
2018-10-17 16:05:48
|
<Thiemo_WMDE>
|
Thanks chiborg!
|
|
2018-10-17 16:11:56
|
<McJohn>
|
👏
|
|
2018-10-17 16:13:13
|
<MGC|mobil_>
|
Thank you!
|
|
2018-10-17 16:14:38
|
<MGC|mobil_>
|
Even if the meeting is over now, how exactly do you do depending patches like that? I never really figured it out...
|
|
2018-10-17 18:26:42
|
<Nemo_bis>
|
MGC: just a series of patches in the same branch?
|
|
2018-10-17 20:31:28
|
<geniice>
|
what's up with the image server?
|
|
2018-10-17 20:32:57
|
<bawolff>
|
Something wrong with it?
|
|
2018-10-17 20:40:59
|
<geniice>
|
Images not showing up for me on commons. However It might be this rather questionable hotel Wifi I'm using
|
|
2018-10-17 20:43:16
|
<bawolff>
|
I don't see any mention of problems on server admin log
|
|
2018-10-17 20:43:28
|
<geniice>
|
fair enough
|
|
2018-10-17 20:43:55
|
<bawolff>
|
geniice: if you got directly to https://upload.wikimedia.org/wikipedia/commons/thumb/e/ec/Sphinx_at_Universitetskaya_Embankment_%28img1%29.jpg/450px-Sphinx_at_Universitetskaya_Embankment_%28img1%29.jpg do you see the picture, does it time out? do you get some sort of error
|
|
2018-10-17 20:48:16
|
<geniice>
|
I get a picture
|
|
2018-10-17 20:48:22
|
<geniice>
|
https://commons.wikimedia.org/w/index.php?search=vega&title=Special%3ASearch&go=Go however throws up an error
|
|
2018-10-17 20:48:37
|
<geniice>
|
Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes.
|
|
2018-10-17 20:48:39
|
<geniice>
|
See the error message at the bottom of this page for more information.
|
