User Details
- User Since
- Oct 10 2014, 8:08 AM (522 w, 1 d)
- Availability
- Available
- IRC Nick
- dues, duesen
- LDAP User
- Daniel Kinzler
- MediaWiki User
- DKinzler (WMF) [ Global Accounts ]
Wed, Oct 2
Mon, Sep 30
Here we see both forms of module specification. The specs/v0 bit actually specifies the specs/v0 module, per the specs.v0.json module definition file. The content.v1 bit would be our new format for specifying module ids as parameters in urls.
But that approach is a bit hacky, because it makes assumptions about the module id structure, and it really only works if the module id is the last parameter. Which means that any endpoint that takes two module ids is problematic.
Thu, Sep 26
Tue, Sep 24
@Mooeypoo Amir just told me about this task. We should definitly re-think at what point events get inserted into the database.
Tue, Sep 17
But it was overriden in CirrusSearch ( https://gerrit.wikimedia.org/r/c/mediawiki/extensions/CirrusSearch/+/1072783 ) which is Wikimedia maintained code and is not the repo that defines it?
Mon, Sep 16
I note that the first of the four listed endpoints ("Get revision metadata for a title") does not yet produce equivalent or compatible output.
Thu, Sep 12
@akosiaris Did I understand correctly that the rewrite rule for /api/ has been rolld back, so this is solved (for now)?
Sep 11 2024
Sep 9 2024
Sep 5 2024
Sep 4 2024
Sep 3 2024
Sep 2 2024
It looks like the /html/to/wikitext and /wikitext/to/html endpoints will stay for the foreseeable future.
Aug 31 2024
Aug 30 2024
In T370304#10073278, @TheDJ wrote:
It used to be that every connection would not return until almost all the work was done, so there used to be a bit of a natural brake on streams like this, but nowadays there is a lot more work that happens after the request completes (for very good reasons btw).
Aug 29 2024
Aug 28 2024
Yesterday we discussed the problems that arise when check can't store the rendered formula indefinitely. In that case, any requests that try to retrieve the rendered formula based on the hash will fail.
Aug 27 2024
Aug 26 2024
Tagging MW-Interfaces-Team for awareness.
Aug 21 2024
Aug 20 2024
Looking at metrics from LinksUpdate, it seems that we could reduce the number of purges a lot, if we wait until after we re-parse to decide whetehr we need to purge or not.
Aug 19 2024
Aug 12 2024
Aug 8 2024
I don't quite understand the reasoning behind this. I think the basic question is: what code should be directly writing to the output buffer (and the WebResponse object)? Should skins build strings, and hand them off to core for output? Or should it be the other way around - core (and extensions) build a data structuer, and the skin generates the response?
Aug 6 2024
Aug 5 2024
Jul 18 2024
Everyone seems to be happy with content.v1, so let's do it!
Jul 17 2024
Jul 16 2024
Jul 13 2024
Jul 12 2024
Jul 11 2024
This is now live for testing on https://test.wikipedia.org/wiki/Special:RestSandbox. Please keep in mind that the swagger specs for the MW REST APIs are still incomplete. This is just a preview.
The root problem is that the fact that the Authority methods need a good Status object as an input isn't really documented.
Jul 10 2024
If I understood that correctly, then that seems reasonable and I have no objection.
Jul 9 2024
Thank you!
I have been wondering if we shoudl take a different approch to this.
We now have the OPT_ENFORCE_JSON_TYPES option, that forces strict type checks on parameters comng from JSON.
We could use that to require an array when PARAM_ISMULTI is set. That way, we'd keep support for PARAM_ISMULTI for body params, without the awkwardness of automatically pslitting strings contained in a JSON structure. For form data, we'd still support that behavior, as before.
Jul 8 2024
Jul 6 2024
Jul 5 2024
This patch should cause the name of the extra field to be included in the hint.