Just use Flow. LQT is severely buggy due to its reliance on wiki pages for message storage. It resulted in a very deep hooking into the MediaWiki internals, causing endless breakages. On Fri, Jan 30, 2015 at 6:13 AM, Jason Ji <jason.y.ji(a)gmail.com> wrote: -- Best regards, Max Semenik ([[User:MaxSem]])

Hi Max and David, Thanks for your feedback. To clarify a bit, we're not thinking of using LiquidThreads as it is - we have a different extension we will be building, with some different needs than LQT has. For example, we may not need any integration with watchlists. So our thought is that we might fork LQT and modify it to suit our needs. We're still very early in the design phase. Max - when you say just use Flow, do you mean we should fork the Flow code base and work from there, or that we should just install Flow? Flow looks interesting, but we're not sure it will have the features we need, and our timeframe is likely to be shorter than the timeframe of Flow development. Is there somewhere I can go read in detail about the bugs and unfixable problems with LQT? We might not fork LQT at all, but we were also thinking of using wiki pages to store comment text. So if that idea is fundamentally broken, it would be great to know why. Thanks! Jason On Fri, Jan 30, 2015 at 12:55 PM, David Gerard <dgerard(a)gmail.com> wrote:

Hi Max, I'm looking at Flow today. The documentation talks about how to replace individual pages or entire namespaces with flow boards (using $wgFlowOccupyPages and $wgFlowOccupyNamespaces). However, is there a way to embed a Flow board at the base of a wiki page, as a more traditional commenting system might look like? Also, is there a way to query for Flow comments with parameters such as Flow comments by user, Flow comments by associated page, etc? Thanks, Jason Ji On Fri, Jan 30, 2015 at 7:40 PM, Max Semenik <maxsem.wiki(a)gmail.com> wrote:

Hi Nick, Thanks for the rapid response! I'm not sure how much I'm allowed to discuss details about our needs (we are working on a project for a client), but my understanding is that it involves a commenting system built on top of our wikis which is 1) easy to use and similar to Facebook, YouTube, or other systems where you can immediately and easily see comments at the bottom of a post or a page, and 2) queryable, perhaps from an API or some other method, such that we could potentially build an extension which could aggregate comments by user or associated page (similar to Reddit, where you can see a user's comments from their user page). In the future, we'd also like to be able to assign high level 'topics' to wiki pages and be able to query for discussion threads related to these topics. Our wikis are backed by Semantic MediaWiki, so we were thinking of using semantic properties to help with the querying aspect if we built our own commenting system, but we're also investigating Flow to see how well it could meet these needs. I'm not sure that the header area of the Flow board is useful to us in the pursuit of our first use case, unfortunately. It sounds like that could be good if we were building a new wiki in which every single page was a Flow board, and the header area was the actual article itself. That would simulate a comment area beneath a wiki page. Unfortunately, we already have an existing wiki with content. Jason On Wed, Feb 4, 2015 at 4:54 PM, Nick Wilson (Quiddity) < nwilson(a)wikimedia.org&gt; wrote:

On 02/04/2015 08:59 PM, Nick Wilson (Quiddity) wrote: Aggregating comments by user in the API is relevant to our interests, I think. To the extent that is not currently possible (there appear to be some limitations in our API support, e.g. https://phabricator.wikimedia.org/T88753 ), we are interested in fixing it at some point. It is already possible to see Flow contributions as part of the Contributions page (https://www.mediawiki.org/wiki/Special:Contributions/Mattflaschen_%28WMF%29). It is already possible to get all comments for a page from the API (in fact, this is the default with view-topiclist. (Topics are paged for performance reasons, so for long boards, you may have to iterate). As Nick said, we have discussed this for our longer-term roadmap. Our implementation would probably not use SMW. Whether Flow will support SMW is also an open issue (we have at least one other interested wiki that uses SMW). In our long-term roadmap, we have discussed integrating the Flow comments into the main article. I don't recommend the header approach you mentioned either. Among other problems, tools that support pages (e.g. VisualEditor) will not be able to work with it properly. (Even when we support VE for editing headers, it will not be the same form of it as used for editing pages). Matt Flaschen