The past, present, and future of UI at GitHub
hawksley.org
hawksley.org
That number seemed surprising to me - I actually expected the portion of mobile-only internet users would be significantly higher than that.
Turns out the 15% number means something slightly different. From that Pew Research page:
"Today, 15% of U.S. adults are “smartphone-only” internet users – meaning they own a smartphone but say they do not subscribe to a home broadband service."
So the 15% is people who use the internet exclusively via LTE/5G without paying for home broadband.
(I'm surprised that number isn't higher as well.)
Does that factor in people “borrowing” their neighbors WiFi?
Gen Z is likely higher. Older generations will still use a laptop.
In lower income countries, this number is also higher. Many people do their entire job on their phone. Email is not used often for business. It's chat apps.
I wish Codespaces worked better on mobile! It would be great to see them invest more work in that.
Anecdotally I've heard that there are people out there who have successfully learned to program entirely on mobile devices.
Occasionally I'll ship feature changes to my JavaScript apps, usually copy and pasted over from Claude.
I reported it in those offline "please rate the new ui" feedback popups, but it's clear that they have no idea who use their UIs.
Nowadays stuff like this is a sea of divs with an ocean of margins and padding. Or just straight up card-based layouts.
> These unique kinds of UIs have less convention and unique accessibility characteristics that are expensive to solve. But a lot of the time, our budget should be zero. To build it with what we have already. To copy-paste.
If you build UIs like this, you end up with the sort of one-size-fits-all UIs no one likes. Like when Twitter (and others) consolidated their desktop and mobile UIs into one mobile-first one that works worse on. You waste a bunch of space and information density, lose eye scanning ability, for the benefit of higher developer velocity. It must be carefully applied. Things like shipping address forms (example from article) or yes/no modals are a great place for standardization but trying to assemble those pieces on a single page, or show something slightly different and you're back to square one.
---
Didn't we try to standardize some of this already? Isnt this why <input>s are derived from the OS and why alert() and prompt() exist? And default style sheets? And no one wanted this! Browsers have opened up _everything_ to be customized, capitulating to designers. And even with all that customization, some things like HTML5 validation still feel like theyre missing pieces.
It should be easy to make things “look like GitHub”, but still possible to build new outside-the-box experiences where relevant.
Projects is a mixed bag, but not terrible. I feel like Github is generally good at making things as complex as they need to be, and no more.
Now? Whatever UI update they've just pushed isn't reflected in the documentation. I can't think of anything that's been improved since GH decided to reinvent text rendering (and poorly at that) and then double down on it.
Upon further inspection it sort of works if I log in. It behaves differently at least. Ugh.
They even completely break super basic things that used to work fine.
For example, if you want the direct download url of (say) an image file in your repo you used to be able to hover over the "Download raw file" button and the status bar would show you the url.
But no, no any more. It's now some kind of javascript button bullshit and there's no way to get it to show the actual download url. Clicking it at least downloads the file, but the history entry is messed up "blob:..." and also doesn't show the download url (using Firefox).
Because at GitHub they clearly can't leave alone the things that actually work, and need to change things just for the sake of change. Seriously not impressed. :( :( :(
Also shouldn't they be able to just know this based on traffic? Why the need to use third-party general research as a source rather than "Our traffic patterns suggest X"?
The one was a crap console all around, but Microsoft’s auth BS is what finally got both of them stuck in our attic. More recently, this happened with minecraft.
GitHub recently forced me to turn on 2FA over SMS. Now they’re running a banner telling me that it has all the problems that were the reasons I didn’t want to turn it on, and telling me to turn on some other 2FA instead.
Pre-software there are plenty of "design of everyday things" books, schools of thought, movements, etc.
This is nothing new, aside from our current state being driven by a recent surge.
People still do their own things of course. Like how tons of Windows apps build completely custom stuff (e.g. Winamp). Or very nearly every single videogame.
> 15 percent of U.S. adults only access the internet through a mobile device.
Absolutely nobody is using GitHub exclusively from their phone. As an aside, this site doesn’t work on mobile.
> But there is a bigger problem. Why are there so many design systems in the first place?
Because homogeneity is boring. I read the article linked about the global design system:
> Many — or even most! — web developers shouldn’t need to understand many close-to-the-metal HTML concepts in order to make web applications function
Next they’re gonna say that web developers shouldn’t need to know how to code. HTML isn’t low level.
> For the love of God, why is there more than one way for me to provide my shipping address on the internet?
For fun? The screenshot shows 3 perfectly good looking forms. I have never ONCE struggled to input my address into a form.
As an OSS maintainer, I do quite a lot of issue triage and simple PR review via the GitHub mobile app, which has gotten better by leaps and bounds in the past 3 years.
> I have never ONCE struggled to input my address into a form.
The different ways forms treat secondary addresses can be a real pain, if you have one of those.
https://github.com/orgs/community/discussions/135572
I don't regret posting my comment (on Sept 4, not shown by default) because it truly was a shit change, but finding a viable alternative to GitHib has definitely elevated in my "get around to it" pile as the preferable alternative to ever giving feedback again. It just exposes how much else interferes with getting things done.How do I make the comments stop? It's noise at this point. My comment isn't in my own account page under "contribution activity". There's no way to search for the comment in the actual discussion thread and trying to search page source for e.g. the username doesn't do any good because not all comments are shown by default; changing the sort to "top" doesn't work because, even though I only got 2 upvotes (and 1 emoji), the initial display still hides the majority (221 out of 251) of items to display lower priority stuff at the bottom... and besides it's broken and threads aren't sorted strictly according to any interpretation of "top" I can muster.
How freaking bold this mofo is! I stopped using GitSht because after being bought by MS it went downhill progressively rendering every crucial feature unusable! Like on purpose, simplest things like buttons and menus went more and more broken. Month by month SOMETHING new went wrong. How f* sinister can devs be to enjoy torturing users so much? Oh, right: this is a corporate policy.
I have 3 browsers that are possible to run on my system and NONE of them fully works with GitHub. When even on PaleMoon issue comments started glitching I gave up on GH. F*k it. I'd rather run my own.
The lower I went to the ground, e.g. backend, then native code, then C... the happier I am. It's like the ocean, the lower you go, the more stable and calm things are. On the surface, you get the crashing waves.
I used to be a Web expert, from the early days, DHTML and flash, to the median years, jquery, backbone etc. to then writing my own libraries. Vue 2 was the last I really used as front-end heavy.
Since then I've been secluded more towards the backend, I leave the front end to the young. I do like systems like Laravel Livewire though, it's mildly sane.
When I write websites now, I do it all by hand and focus on really fast page loading times rather than a monstrosity of generated code and async over the wire. I have such a better time doing it.
Today's web should be considered harmful to the health. It's a high blood pressure job.
Trying to fit my application in 1kB of flash and one hundred twenty-eight entire bytes of RAM taught me a lot as a young programmer. It really makes you confront the physicality of the machine your code represents. I don't think nearly enough programmers have that experience
This is such a bullshit take. I don't think I ever used my phone to do anything on Github. Nor do I want to.
Do they really want to move users off Azure DevOps to GitHub?
Asking as someone about to invest in AZDO.
Which probably doesn't bode well for AZDO as it's not a definite no, but on the other hand it's not clearcut that they are actually pushing that.
The general consensus among the great unwashed is that AZDO is turning into an undead project that won't improve. But the reality is there's a lot of present enterprise investment in AZDO and no clear migration path, as well as a lot of internal MS use of AZDO, so it's doubtful it's true but more a loud PM in GitHub over egging his own product.
But it's sort of turning into a self-fullfilling prophecy as people are getting skittish about investing in AZDO.
The reality is that unless you have crazy builds switching is a few days work and more a pain in the ass than a real project.
The design is such that even if you have 100 active repositories, it is easy to use.
GitHub has no such vision. The interface is clean and easy to use, but the depth of customisation is nearly nonexistent. And the interface keeps bringing you back to the public area of GitHub.
Would be very disappointing if they killed off AZDO. I don’t think GH is an alternative, at all.
1) A UI that _cleanly_ shows me things assigned to me and opened by me
2) Filter PRs by directory.
Phabricator did both of these quite well.
Will there ever be a way to dispose of the enormous header blocking me from reading source code?
Those are what I want from the future of UI at GitHub.
Github is a lot of things to many people. A "web host" and a way of publishing information about software is definitely one of the things it is.
Most software is not popular opensource with a carefully curated web footprint.
(What would suit me, I think, is a view that was very much like the current one, but with the files list collapsed (but expandable). So there'd still be all the the tab-like row across the top, and the About column with releases link, and the main area gewgaws like branch dropdown and some indication of the most recent commit. But just with the README front and centre by default.)
But it's mostly just a sign of how almost nobody actually cares about the code now, and just use GitHub as a source of free libraries that may as well not have source available at all.
For others of us, a README is something we may briefly read once or twice when first considering or integrating a project, but the repo's code is where we spend way more time -- evaluating quality, tracing potential bugs, unearhing quirks and overcoming sparse documentation, considering whether a fork is needed, etc
But that's all hokey old greybeard stuff these days.
This website is a step back. One cannot say if there are some slides (presentation) with handouts (??) or something else.
> Which is all to say, that mobile is the new baseline.
I'm sure a lot of mobile users browse github. Phones and tablets are a wonderful development environment. Notice the diversity of code editors and debug tools running on the phones and tablets. /s