Our great sponsors
-
oil
Oils is our upgrade path from bash to a better language and runtime. It's also for Python and JavaScript users who avoid shell!
-
-
WorkOS
The modern identity platform for B2B SaaS. The APIs are flexible and easy-to-use, supporting authentication, user identity, and complex enterprise features like SSO and SCIM provisioning.
Paul Graham has an essay somewhere where he describes Robert Morris (creator of the Morris worm, kernel hacker, professor at MIT) as the FreeBSD sys admin for ViaWeb :) That is, probably the "most overqualified" sys admin in the world, because he was owning the whole problem.
----
The OP's post is related to the motivation for my work on https://www.oilshell.org/
> I had one developer take 6 months to build a (relatively simple) top nav for a web app. This shouldn't have taken more than 1-2 weeks, even with a careful eye for detail.
Oh, you mean "bikeshedding."
Here's an example of the difference between basic quality, and High Quality:
If you look at most of the repos for SPM modules in my portfolio[0], you'll see that the vast majority have test harnesses. I prefer using test harnesses[1].
These test harnesses tend to be pretty damn robust apps. Many are "ready for app store" robust. A lot of folks would just publish them, "as is." I've been writing apps for a very long time. I'm fairly good at this.
I can write a fairly good test harness, with full app capabilities, in less than a day. If I take the time to localize it, maybe add a day or so.
Here's an example of some test harnesses[2]. Note that there are four of them. These represent the four different target environments for Apple (iOS/iPadOS, WatchOS, TVOS, and MacOS). I'll probably need to fork iOS and iPadOS, in the future, but we're not there, yet. A single codebase is still good for both.
They test a Bluetooth framework[3].
It probably took me around a week or so, to write each one. They are pretty damn good. I think they are all "App Store ready."
I decided to actually go ahead, and create a set of apps, based on these[4], [5], [6].
I spent well over a month, on each, after merging over the test harness codebases, to make them ready for the App Store. Lots of UX testing, removing code that only applied to testing, and adding "friendlier" user interface.
I'm working on an app that I started about a year ago. Actually, I started it over ten years ago, if you include the two servers that I wrote, upon which it depends.
One of the reasons that it has taken so long, is that I have truncated months of work, and tossed them in the garbage, because they were not the proper way to go. I have an "evolutionary design" process[7], that means this can happen. I plan for it. I've probably shitcanned three months' of work.
Another thing that I do, is have an "always beta" approach to Quality. I maintain the product at "incomplete, but ship Quality" status for as much of the project as possible. In fact, I've been sharing it with the team, using TestFlight, since Oct 3, 2020 at 7:47 AM (I got that from the TestFlight metadata).
That means that the app has been stable and robust enough for user testing, and approval for basic App Store release (TestFlight External Testing is a more relaxed standard, but try pushing out a crasher, and see how far that goes).
I add localization support, accessibility, Dark Mode support, leak testing, etc., at every turn. It's very useful, because I can solicit immediate feedback from non-tech team members. It also means that the "basics" for App Store release are constantly being tested and validated.
Even more useful, if we want to ask for money, it's dam easy. We just loop the person we're begging from, into the TestFlight External Tester pool, and they can run the app without a Marketing chaperone, or sacrifices to the demo gods. We can also get valuable feedback from them.
It's really, really nice, and it has been, for many months.
I feel like we are now at a "starting point." Even though it has been a fully-functioning, release-ready app for the last couple of months, it need the "MVP treatment," where the testing pool is expanded, and we start applying it to "in the wild" scenarios.
Lots of companies use their customers as guinea pigs for the first several releases; usually by shoving baling-wire-and-duct-tape junk down their throats (and making them pay for it), before hitting their stride. It's a deliberate strategy. Some months ago, I read a post, here by a founder, declaring that "if you don't get physically sick at the quality of the code in your MVP, then you are spending too much time on the code quality."
Basically, deliberately write garbage, and force it on your users.
One of the reasons that I took on this project, was the founder is a friend of mine. He is running it as an NPO (501c3), and putting his own money into it. He doesn't really have much of it, to begin with. Also, more alarmingly, he didn't actually have a particularly good idea of what, exactly, he wanted the app to be. That's a recipe for disaster.
He asked me to help him vet some development shops he was approaching, to realize his vision.
It was eye-opening. He got a number of ridiculous quotes. I know what is necessary for this type of project (not small). For example, when one said that they'll deliver a full multi-server, multi-client app for MVP in three months (firm), upon getting a vague, hand-wavy requirements spec, it was hard for me to keep a straight face.
After a few of these, I just got disgusted, and said "Screw this. I'll do it." I've been developing it for free, as a native iOS/iPadOS app.
He has to pinch himself.
[0] https://stackoverflow.com/story/chrismarshall
[1] https://littlegreenviper.com/miscellany/testing-harness-vs-u...
[2] https://github.com/RiftValleySoftware/RVS_BlueThoth/tree/mas...
[3] https://github.com/RiftValleySoftware/RVS_BlueThoth
[4] https://apps.apple.com/us/app/blue-van-clef-for-mobile/id151... (iOS -Includes Watch app)
[5] https://apps.apple.com/us/app/blue-van-clef/id1529005127 (Mac)
[6] https://apps.apple.com/us/app/blue-van-clef-for-tv/id1529181... (TV)
[7] https://littlegreenviper.com/miscellany/evolutionary-design-...