Fri, Sep 22
Minified JS is not binary code.
Thu, Sep 21
Is it still observable in modern MW/HHVM versions?
Wed, Sep 20
This probably needs to be closed due to T176370: Migrate to PHP 7 in WMF production - unless someone thinks that HHVM's PHP7 mode can be a convenient way to test stuff before PHP7 migration proper (which I strongly doubt).
Not happening due to T176370: Migrate to PHP 7 in WMF production.
Tue, Sep 19
Mon, Sep 18
Which actually raises the question of when we'll have to make a decision between Zend and HHVM.
Fri, Sep 15
The above pull request has been merged, now tilerator needs a deploy. Looks like I'm still the only person at WMF who has ever deployed the maps services - I'd rather have someone else try it now that there's me in addition to the docs I wrote than wait for me to get hit by a bus.
Thu, Sep 14
Ugh, that patch is a copyright violation itself: it assigns a GPL header (and a wrong version of GPL, not used by MW) to code that may come under a different license (for example, we use jQuery which comes under MIT), and assigns it to an nonexistent "Mediawiki Foundation", ignoring real authors' copyrights.
Wed, Sep 13
This should definitely be a new version, 1.5.
Tue, Sep 12
Fri, Sep 8
Dunno, I've seen it used as A in various leet renderings, but how high is the actual chance someone might not notice the difference?
Thu, Sep 7
I might have time to look at this next week.
Mean time is meaningless, I'd rather see p75, p90 and p99. I've made some changes, please check out.
Tue, Sep 5
I mean there should be a variable for that.
For that particular page.
We can just locally disable VE on eligible pages for eligible users.
Still EventLogging =)
Sat, Sep 2
Fri, Sep 1
Thu, Aug 31
Wed, Aug 30
Can I get the usernames of accounts that had received such messages when they shouldn't have?
Updated the page. Basically, redirecting you to the special page and displaying whatever article-creation-landing-page contains on the local wiki is all the functionality of this extension:)
Tue, Aug 29
The patch implementing this has been merged, so at this point I'm confused what's the point of continuing to go through this formal process.
Mon, Aug 28
Appears fixed for me. Please reopen if I'm wrong:)
Fri, Aug 25
Aug 24 2017
Current situation: we create jobs, nothing from jobs ends up in the log:
Aug 23 2017
@Florian, do you still need this?
Sounds like a plan!
@kaldari, the core does that for us, CentralIdLookup falls back on local IDs without CentralAuth.
Aug 22 2017
Besides "loss of entropy" being a misnomer (see e.g. here), we use CSPRNG mostly for session encryption and the typical number is under 2 logins per second. Which means that overall CSPRNG usage across the cluster is negligibly low and we're way too far from "exhaustion" of any kind on this front.
Aug 21 2017
In principle, nothing prevents us from switching deployment-prep right now by editing https://wikitech.wikimedia.org/wiki/Hiera:Deployment-prep I think that would be something like
We may do so, however the usefulness if this would be fairly limited because we will not be able to use any 7.0 features anyway until we bump the MW requirements.
I've created a basic fuzz test at https://gerrit.wikimedia.org/r/#/c/372870/ - if you could add some fuzzing targeted at your new code, I would be more confident about +2ing this project.