(2/2)
@jupiter_rowland
> they're famous for having separate repositories for the server and the Web frontend
You say this like it's a bad thing. They are separate pieces of software, with different purposes, requiring totally different skillsets. Treating them as such is exactly what we need more of.
Wouldn't Mastodon would be better if it specialised in developing apps, and outsourced the server side to people who know how to do back-end engineering?
You say this like it's a bad thing.
Not at all.
One advantage is, as you've said, that the backend and the Web frontend can have their own developers, development of both can largely be detached, and they can be upgraded separately from one another.
Separate Web frontends can be developed by people who actually know a thing or two about frontend development and UI design. I mean, look at the Web UIs of some all-in-one Fediverse server applications. They're often the digital counterpart of random knobs and switches poked through a piece of cardboard and labelled with a ball pen, just so that these knobs and switches are there. Sometimes they're the equivalent of expecting all kinds of end users to operate DIP switches, but hey, they're still better than soldering and unsoldering wires.
Another advantage is that server software for which alternative frontends exist does not have to drag its default frontend around. There are Mastodon servers with alternative frontends, but they still have to have the two official Web UIs installed (the default one and the Tweetdeck-style one) because they're firmly welded to the backend. I guess we all know what a heavyweight Mastodon is, and I'm certain that part of the weight is caused by the built-in Web UIs. In stark contrast, you can set up an Akkoma server with Mangane instead of Akkoma-FE, as in without having to also install Akkoma-FE.
By the way, Hubzilla is an interesting case here. Not only is its default UI very configurable, but Hubzilla itself is highly themeable, and third-party themes almost amount to entirely new UIs. At the same time, however, practically all official development efforts went only into the backend for most of its existence.
Any Hubzilla UI has to wrestle an immense wealth of features, and not exactly new features were added over time. This, however, caused Hubzilla's UI to gradually turn into a jumbled mess because some of the new UI elements were seemingly added in totally random places. Not only was the UI never cleaned up, but the default theme is perpetually stuck in 2012 (the name "Redbasic" says it all, it was made for Hubzilla when Hubzilla was still named Red), it was derived from an early Friendica theme, and even Friendica wasn't pretty back then. Also, the documentation was completely neglected.
So the situation last year was that there was only one working Hubzilla theme left, and that was Redbasic. It was the only theme that was even only upgraded to work with newer Hubzilla versions. There used to be other official themes, but they eventually ended up so outdated that they were removed altogether. @Sean Tilley's third-party themes were last touched seven years ago, that must have been around the time when Hubzilla 3 came out. At the same time, the official documentation was not only highly incomplete, but it was so outdated that parts of it were simply false. It partly referred to features that had been axed many years ago (tech levels) and features that simply were never there (four different mention styles), and parts of it even still spoke of Red. Thus, nobody even knew how to develop new themes for current Hubzilla.
That was when the community stepped in. @ Der Pepe (Hubzilla) ⁂ ⚝ sat down and rewrote the entire help. @ Scott M. Stolz not only started working on his NeuHub themes, but in the same process, he reverse-engineered Hubzilla's theming system to write documentation for theming Hubzilla which had never been written before AFAIK. Around that time, @ ????? was dabbling with specialised themes for certain purposes, e.g. one very clean theme for Hubzilla channels used as long-form blogs. Later on, @ Saiwal joined the fray with his now-popular Utsukta themes.
Granted, Hubzilla still carries Redbasic around, not only as the default for new channels unless the admin chooses another one, but also as a fallback in case a new Hubzilla version doesn't support existing third-party themes anymore. The latter is becoming less likely as the Utsukta themes are being built against Hubzilla's development versions now. Besides, it's in Hubzilla's nature that everything on a hub is updated at the same time, including third-party repositories.
In general, the Hubzilla community is no longer that easily satisfied with a UI that "just works", and the devs have taken notice. Hubzilla 10.4, now a release candidate, will spruce up certain core parts of the UI. It will introduce a tree-style thread view as the new default instead of its current chronological view, something that Friendica, (streams) and Forte have had for significantly longer. That is, this is actually a side-effect of the introduction of "lazy loading" conversations to reduce the server workload. Also, upon user request, it will add a button to add images to comments.
If (streams) and Forte grow bigger, the same could happen there. They have two official themes to choose from, fairly new Fresh and an older version of Redbasic. However, they don't have a large enough community for all the same things to happen to them that happened to Hubzilla, although Pepe has said he'd rewrite the (streams) and Forte help as well, seeing as Mike had ripped them out entirely with no replacements as they were too outdated at that point. Maybe someone will even write a guide on how to adapt Hubzilla themes to (streams) and Forte.
That is, (streams) and Forte are both already the result of several years of UI and UX advancement and improvements and making them fit for a Mastodon-dominated Fediverse (where Hubzilla is still geared towards a Fediverse which it will dominate itself by the mid-to-late 2010s). This is stuff which can't be taken care of in themes because it concerns the UI engine itself, and it's partly tied deeply into the backend.
While Hubzilla, (streams) and Forte won't be able to do without their official themes anytime soon, the official themes don't significantly weigh them down. Still, they require some maintenance work to keep up with the backend.
Wouldn't Mastodon would be better if it specialised in developing apps, and outsourced the server side to people who know how to do back-end engineering?
This makes me wonder which half Mastodon would be willing to outsource. I think they'd rather hold on to the backend and pass all the frontends on. Of course, this would come with the advantage of the official Mastodon mobile app actually becoming somewhat decent rather than remaining the "we need an official app, no matter how" kluge that it is today.
CC: @Tim Chambers @rakoo
# Long # LongPost # CWLong # CWLongPost # FediMeta # FediverseMeta # CWFediMeta # CWFediverseMeta # Fediverse # ActivityPub # Mastodon # Akkoma # AkkomaFE # Mangane # Akkomane # Hubzilla # Streams # (streams) # Forte