Talk:Matrix (protocol)

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Notability for advancement from Draft:[edit]

I've added two more sources. So now it has citations to independent articles from Dr. Dobbs Journal, LWN.net, itWorld and tvTelecom, along with one other independent cite. They were described by one helpful person on #wikipedia-en-help as 4 "gold-standard-in-depth-sources". They span a period of 9 months. I think that qualifies as wp:notability ★NealMcB★ (talk) 19:05, 20 July 2015 (UTC)[reply]

Logo Size[edit]

It's a bit small, I'm not sure if there is anything that can be done to make it a bit bigger so it fills out the box? 185.112.0.202 (talk) 10:51, 7 September 2018 (UTC)[reply]

I re-uploaded the logo in vectorial format acquired from Matrix.org's Github Page, size set to 220px, which I think is a sensible size. Daimond (talk) 22:11, 10 September 2018 (UTC)[reply]

Information about Matrix.org[edit]

I've added some information about matrix.org's recent breach. While it isn't related to the protocol itself, I think it's worth mentioning as this does affect the matrix ecosystem as a whole.
On a more general note, should information about matrix.org be posted here? I don't think matrix.org is notable enough to receive its own page. GautamC129 (talk) 02:22, 13 April 2019 (UTC)[reply]

AFAIK putting information about the matrix.org breach here is like putting information about the breach of Hilary Clinton's mailserver on the Wikipedia pages about email. I'd say it's out of scope, as the vulnerability that allowed the breach had nothing to do with the Matrix protocol itself. I agree that matrix.org probably isn't notable enough to have it's own page, but information about it is probably better on a page about the New Vector company that runs matrix.org, or perhaps a page about the Matrix Foundation that steward the protocol. --Danylstrype (talk) 12:22, 22 May 2020 (UTC)[reply]

Confusion between Matrix protocol and Riot integrations[edit]

I'm concerned that the current text on this page confuses the Matrix protocol (the topic of the page), and various integrations that the Riot app (the flagship app using Matrix) has with other systems that are not part of the Matrix protocol. For example, Riot integrates with Jitsi Meet to provide audio/ video conferencing, and JM does use WebRTC, but AFAIK WebRTC is not part of the Matrix protocol itself. --Danylstrype (talk) 12:33, 22 May 2020 (UTC)[reply]

Matrix does use WebRTC for 1:1 VoIP/Video calling. Source: https://matrix.org/docs/spec/client_server/r0.6.1#voice-over-ip Marviiin (talk) 22:11, 22 July 2020 (UTC)[reply]

Stylized name[edit]

@Marviiin wanted to remove the stylized [matrix] from the lede, but due to apparent conflict of interest I thought this could use discussion. I see File:Matrix logo.svg uses this stylization. With a quick web search, heiCHAT (Heidelberg University) uses this stylization. I recall seeing the brackets associated elsewhere too, but don't have references available right now. Per MOS:TMSTYLE, I think it's appropriate to keep the stylized version. 84.250.17.211 (talk) 20:20, 22 July 2020 (UTC)[reply]

So, I appreciate I'm a primary source here and thus conflicted, but if you look at the entirety matrix.org site and the vast majority of projects which build on the protocol, they refer to the protocol as Matrix in writing. People sometimes get confused because the logo of the project (i.e. the SVG) happens to have square brackets around the word matrix, but this is just cosmetics - it's *not* the name of the project. Just as you wouldn't try to render their logos of any other companies in unicode (e.g. you wouldn't refer to Apple Inc as  in writing, or Microsoft as ⊞Microsoft). For instance, https://matrix.org/blog/2020/06/16/matrix-decomposition-an-independent-academic-analysis-of-matrix-state-resolution describes an academic paper written about Matrix (which happened to originally get it wrong and use [matrix] to try to disambiguate from the mathematical term Matrix - but was thankfully corrected before it went to press). Marviiin (talk) 21:42, 22 July 2020 (UTC)[reply]
In practice (a quick search on this wiki), there are articles such as McMillions (stylized as McMillion$), Positiv (stylized as POSITiV), Snow (app) (stylized as SNOW), Razer Inc. (stylized as R Λ Z Ξ R), VG247 (stylized as VG24/7), etc. Many of these stylized names come from the logoface, even if the official website refers to them by the unstylized name. 84.250.17.211 (talk) 01:44, 23 July 2020 (UTC)[reply]
To me it looks like a wider issue. No one really uses stylized names. Imho this is the reason to drop it even in articles that have them. Take Razer for example. It's logo is a graphic image so any attempt to translate it into unicode symbols is a WP:OR – I haven't found a source to stylization on any of linked articles (except for capitalization maybe). We have a similar situation here. I'm strongly in favor of dropping the stylization, as it's either not notable or even original research. – K4rolB (talk) 16:02, 23 July 2020 (UTC)[reply]
I think having a transcript of the text in an logo is useful for accessibility (if the transcription is possible: Using Greek letters such as Λ Ξ because their glyphs look like the stylized Latin letters in File:Razer wordmark.svg is a hack and it is not clear whether to use Ʌ or Λ. See also: c:Category:Unusual_letterforms). But this information does not belong in the article if the stylized text is not used in writings. In these cases, transcripts should be added as inscription (P1684) c:Commons:Statements and stylized texts removed from the articles. —Dexxor (talk) 09:13, 24 July 2020 (UTC)[reply]
Additionally, something like Slack (software) doesn't currently mention a stylized name, but the logoface is lowercase slack. 84.250.17.211 (talk) 02:08, 23 July 2020 (UTC)[reply]
Perhaps a solution could be to say "sometimes incorrectly stylised as [matrix]"? Marviiin (talk) 22:12, 22 July 2020 (UTC)[reply]
I don't believe the stylization to be "incorrect", unless there's a reliable non-primary source stating otherwise. (See also my comment above.) 84.250.17.211 (talk) 01:47, 23 July 2020 (UTC)[reply]
If you look at every single non-primary source in the Matrix ecosystem - e.g. every(?) independent project linked from https://matrix.org/docs/projects/try-matrix-now/, you will see that the project is referred to as Matrix. Just because Heidelberg were ill-informed and got it wrong doesn't mean that Wikipedia should perpetuate the bug by the very first sentence of the Wikipedia page enshrining the error :) Marviiin (talk) 09:07, 23 July 2020 (UTC)[reply]
No consensus, from the latest Element release. 84.250.17.211 (talk) 21:32, 24 July 2020 (UTC)[reply]
No, there is perfect consensus. The bad commit you linked to was never merged, caught in review, and subsequently reverted, as per https://github.com/vector-im/riot-web/pull/14566#pullrequestreview-451610320. Meanwhile, the "[matrix]" in welcome.html and the matching i18n is a magic constant which is never shown to the user, which is interpolated to be the actual matrix logo SVG + hyperlink, as you can see if you go to https://app.element.io as a new user. The code for this is at https://github.com/vector-im/riot-web/blob/ffab08bd60be34dd34cd0db7c0aa6eca986ce52e/src/components/structures/VectorEmbeddedPage.js#L32. 92.234.32.44 (talk) —Preceding undated comment added 21:43, 24 July 2020 (UTC)[reply]
VectorEmbeddedPage.js seems to display alt text [matrix], even if what you said is true. I see that's also the case on the Element page you linked to. And yes, I linked to both commits appropriately; thanks for finding and linking to a pull request. The Git log graph actually looks like so: 
*   264bce7c Merge branch 'tmp1' into develop
|\
| * 0ada7b5d Update i18n
| * 3e37f21e Revert "[matrix]" to "Matrix"
| * 96f0d3a0 Capitalize and fix [matrix]
[...]
|/ /
* |   398033e7 Merge pull request #14572 from vector-im/t3chguy/devdep
* |   398033e7 Merge pull request #14572 from vector-im/t3chguy/devdep
|\ \
| |/
|/|
| * d328cab5 Move dev dep to the right place
|/
*   da7550e0 Merge pull request #14552 from vector-im/dependabot/npm_and_yarn/lodash-4.17.19
Both were merged, though that's irrelevant WP:OR here from me. 92.234.32.44 geolocates to the UK. 84.250.17.211 (talk) 22:08, 24 July 2020 (UTC)[reply]
Sorry, 92.234.32.44 is me - didn't notice i'd been logged out. Good catch on the alt tag being incorrect; fixed in https://github.com/vector-im/riot-web/commit/f21352f55817eeee49f392405f00284ea615aef8. The fact that https://github.com/vector-im/riot-web/pull/14566 wasn't squash-merged and so technically the bad commit leaked into the repository before immediately being reverted doesn't change the fact that the commit was wrong, which is why it was reverted and fixed. Please don't let the same bug spread into Wikipedia! Marviiin (talk) 01:30, 25 July 2020 (UTC)[reply]
matrix.to uses [matrix] in title and alt text. As far as I'm aware, this is a primary/secondary source too. Matrix is still dominant on web searches. 84.250.17.211 (talk) 01:17, 26 July 2020 (UTC)[reply]
Oops. Fixed: https://github.com/matrix-org/matrix.to/commit/5fd94e7043a328039c8cc90a7dccd677c73c4314 Marviiin (talk) 09:57, 27 July 2020 (UTC)[reply]
Nobody types [matrix]. We're usually using a contraction [m]. VINTproYKT (talk) 08:28, 9 May 2023 (UTC)[reply]

Okay. So it seems to me to be clear that the primary sources (official website, Element and @Marviiin here) has referenced the unstylized name to be canonical (particularly that GitHub pull request comment & commit). I largely agree with @Dexxor's comment where transcriptions of logos should go. I also agree with @K4rolB's comments that many of those examples I gave could constitute as WP:OR in those articles (except for capitalization). But here, we have an example of primary/secondary source(s) formerly saying [matrix], later changed/fixed to Matrix, and at least one third-party source mentioning [matrix] once. I note generally Wikipedia articles should be built on what third-party sources say about a subject, not what primary sources want to say about themselves; most of the references in the article use Matrix, not [matrix]. Is it appropriate to start a larger WP:RFC for this, or should we consider a third-party source such as Heidelberg University page as unreliable and proceed with removing the stylized name from the article? Alternative options (for an RFC) would be an option to say "sometimes stylized as", or "sometimes incorrectly stylized as". The lattermost statement could be referenced to the self-published GitHub pull request comment, at best (which is also not a reliable reference for this article imo). 84.250.17.211 (talk) 04:40, 25 July 2020 (UTC)[reply]

The least we can do is changing it to “sometimes stylised as …” to make a neutral statement that does not say something about the official spelling. —Dexxor (talk) 08:28, 25 July 2020 (UTC)[reply]
A few more datapoints on this: having actually looked at the Heidelberg page at https://www.urz.uni-heidelberg.de/en/heichat, it turns out they are correctly referring to Matrix as Matrix. The only place they don't is the target of the hyperlink to Matrix.org, which is clearly wrong (given nothing at Matrix.org uses that stylisation). This is very clearly incorrect. As another datapoint, it transpires that someone created a clone of the ~2014 vintage Matrix.org website over at Martix.org (http://web.archive.org/web/20190405072452/http://martix.org/), switching the word Matrix for Martix everywhere, for reasons unknown. Just because someone does something weird like this on the internet and so propagates an error, does that mean that Martix is an acceptable stylisation of Matrix? Surely not. Similarly, if someone starts misspelling Matrix with square brackets because they're confused by the logo, it's a bug. I appreciate the sensitivities of WP reflecting the reality of the world rather than whatever an original source might want to be reflected, but in this instance, I really suggest looking through all the hundreds of 3rd party projects building on Matrix at https://matrix.org/docs/projects/try-matrix-now/. None (or almost none) of them use the incorrect stylising. Likewise none of the press about Matrix linked in the article use the incorrect stylising. Therefore, a handful of typos on the internet should surely not be arrogated to the first sentence of a wikipedia article. It'd be like saying "Microsoft, sometimes stylised Micorsoft", just because Google returns 975K hits for the typo Micorsoft (https://www.google.com/search?q=%22micorsoft%22). Marviiin (talk) 00:25, 26 July 2020 (UTC)[reply]
plus Added sometimes, seems sensible and neutral. I found it a bit difficult to propose a WP:RFCNEUTRAL question here. I'm not opposed to starting an RFC, so I've left the {{Disputed inline}} there for now. A proposed neutral question (without options to choose from) for an RFC could be: Should this article's lede mention [matrix]?, or perhaps less confusingly, Do you support the inclusion of [matrix] as a stylized name in this article? 84.250.17.211 (talk) 01:30, 26 July 2020 (UTC)[reply]

Date of introduction[edit]

Any reliable reference for the date of introduction? Special:Diff/897572769 changed it to be different than what the reference stated; I've tagged {{Failed verification}} appropriately. Pinging @Marviiin if he knows more. 84.250.17.211 (talk) 04:47, 25 July 2020 (UTC)[reply]

Sept 3rd 2014, as per https://matrix.org/blog/2014/09/03/welcome-to-the-matrix-blog, also corroborated via https://www.computerworld.com/article/2694500/matrix-wants-to-smash-the-walled-gardens-of-messaging.html Marviiin (talk) 15:03, 2 August 2020 (UTC)[reply]

Criticism?[edit]

Maybe this article should tell something about the performance and scalability issues of the current implementations of Matrix, but I am not sure if there are much or any good, reliable sources about that. --ilmaisin (talk) 21:25, 26 October 2020 (UTC)[reply]

Performance issues and such are almost entirely attributable to synapse, Matrix's "reference implementation", I don't think it's a good idea to blame the protocol for these performance problems, especially when (experimental) homeservers like conduit and dendrite show that significant speedups compared to synapse.
That said, if you can find articles pointing out slowdowns attributable to the core protocol (per prospect of Computer Science concepts such as Big O Notation which conceptualise the efficiency of an algorithm), feel free. Shadowjonathan (talk) 23:53, 29 January 2021 (UTC)[reply]
I think that criticism of the reference implementation can be included in the article if it is well-sourced and described exactly as that, because at the moment it is the most common server implementation and the default experience for the large number of Matrix users. — K4rolB (talk) 10:29, 30 January 2021 (UTC)[reply]
For future reference: Synapse is not any more considered the "reference implementation"[1]. It is just one of the multiple available servers now. MTRNord (talk) 15:33, 1 March 2022 (UTC)[reply]

References

  1. ^ https://github.com/matrix-org/synapse/commit/6744273f0b92457247166299981d69e9f4219601

Support in Thunderbird[edit]

Perhaps for the "History" section? From Thunderbird team outlines plans for the email client:

'"version 91 ... scheduled ... the third quarter of 2021... Introduce Matrix chat support"'

--Mortense (talk) 21:08, 6 June 2021 (UTC)[reply]

Does not seem notable imho. It is only noted in other changes and I suppose nothing came out of it. –K4rolB (talk) 21:13, 8 June 2021 (UTC)[reply]

Directed acyclic graph[edit]

@David Eppstein: I'm not sure as to the best way to include this in the article prose (which is in rather poor shape) but the underlying Matrix protocol is a directed acyclic graph as described in the Matrix spec here. Elli (talk | contribs) 19:34, 2 October 2022 (UTC)[reply]

It USES directed acyclic graphs, and more specifically some kind of event graph. That does not make it an EXAMPLE of a directed acyclic graph, or an article about the theory of directed acyclic graphs. If we had a separate article about acyclic event graphs, it could go in the DAG category, but putting this article in that category is unhelpful to readers because they will not learn anything about directed acyclic graphs by reading the article and will just be confused about why it is even listed in the category. —David Eppstein (talk) 19:37, 2 October 2022 (UTC)[reply]
Perhaps we should have a category such as "Applications of directed acyclic graphs"? Elli (talk | contribs) 20:44, 2 October 2022 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Matrix_(protocol)&oldid=1208374127"