5 ms·
I think their bigger problem is there's shit for documentation. You get a big list of function signatures and if you actually want to know how you're supposed t
by wlesieutre 1y ago
I think their bigger problem is there's shit for documentation. You get a big list of function signatures and if you actually want to know how you're supposed to use anything you find the WWDC session from 4 years ago and hope it's still accurate.
- xcf_seetan 1y agoSomebody told me that the quality of a company is directly proportional to the quality of its documentation...
- thfuran 1y agoI guess mathworks is one of the best companies ever.
- makeitdouble 1y agoBoeing's documentation is probably excellent. In general it is assumed that documentation is the boring part and paying attention to it is a sign of quality. But where do you put people who prefer writing and/or teaching to thinking long and deep about product issues ?
- reactordev 1y agoYou encourage them to blog about it and you aggregate the posts into documentation - somehow. (I'm sure there's people for that, or an AI, or something). They had to explain it when they implemented it - explain it on a post. Confluence. Somewhere. You're right though.
- unscaled 1y agoPerhaps the biggest issue with the 737 MAX was lack of documentation. Boeing consciously chose to completely omit several new features (most notoriously MCAS) from the pilot manuals in order to make it seems like nothing important was changed that would require retraining 737 NG pilots on a simulator. In a sense, this is far worse than what most tech companies do since it is a life-and-death issue and Boeing has made a very conscious effort to hide details here, rather than just being lazy.
- cpgxiii 1y agoOne could say that "the point" of MCAS was to be an implementation detail, something that you would often deliberately hide from software documentation so that users don't design around internal details. There's something of a history of aerospace vendors omitting "implementation details" that end up contributing to serious accidents (e.g. if you get an Airbus far enough out of the normal envelope protections, you lose stall warning), and an equally sordid history of flight and maintenance crews improvising procedures to the observed (rather than designed/specified) behavior of aircraft systems. Arguably, the single biggest systematic risk in the current pilot training system is that crews overlearn to the implementation details of their training, rather than the actual principles and flight manuals (e.g. training inadvertently training for quick engine shutdowns, when the consequences of shutting down the wrong engine in reality are much more serious).
- sofixa 1y ago> There's something of a history of aerospace vendors omitting "implementation details" that end up contributing to serious accidents (e.g. if you get an Airbus far enough out of the normal envelope protections, you lose stall warning), And Airbus control laws and protections are well defined and studied by pilots training for them.
- cpgxiii 1y agoThey are now, just as MCAS behavior is well-defined and covered in conversion training now. They were not covered as well prior to AF447.
- eru 1y agoTech writer. Google famously employs quite a few of them for internal documentation.
- serf 1y ago>Boeing's documentation is probably excellent. it isn't, it's just ISO/NADCAP conforming.
- jiggawatts 1y agoI've read one of the manuals for the 787 management software stack, and it was... beautiful. It read like a novel. I read it twice on the train for the simple pleasure of it.
- dlivingston 1y agoMicrosoft has excellent documentation. So that's one counterexample. :)
- cosmotic 1y agoIt was excellent. It's a mess last time I looked.
- userbinator 1y agoTheir documentation for the very newest stuff contains very obviously AI-generated inaccurate and misleading crap.
- userbinator 1y agoDon't believe me? Go look at it yourself. They've even "edited" the old docs with AI slop. https://learn.microsoft.com/en-us/windows-hardware/drivers/usbcon/transfer-data-to-isochronous-endpoints https://learn.microsoft.com/en-us/windows-hardware/drivers/u... "The characteristics of the endpoint determine the size of each packet is fixed and determined by the characteristics of the endpoint."
- WithinReason 1y agoIf they used AI the sentence would be coherent
- noname120 1y agoThat sentence was already there _at least_ since Dec 2022, stop seeing AI everywhere. https://web.archive.org/web/20221214171028/https://learn.microsoft.com/en-us/windows-hardware/drivers/usbcon/transfer-data-to-isochronous-endpoints https://web.archive.org/web/20221214171028/https://learn.mic...
- userbinator 1y agoThey were using an automated editing tool before that.
- internetter 1y agoA WWDC session, I might add, with a mildly relevant title and rarely a hyperlink from the relevant documentation, let alone a timestamp.
- lilyball 1y agoBut with a full searchable transcript
- internetter 1y agoMachine generated transcript*
- grishka 1y agoBut if you find anything with the dark-blue "documentation archive" header, then you know it's some good stuff. Though sometimes outdated.
- trinix912 1y agoWe've come from the monstrous Inside Macintosh tomes, to that early OS X documentation (that's now in the documentation archive), to this seemingly programmatically generated from doc comments thing.
- grishka 1y agoNext they'll start generating documentation from code using LLMs...
- DaiPlusPlus 1y agoThat's an odd way to spell "AI hallucination"
- saagarjha 1y agoMost of this stuff uses APIs greater than four years old, so you're probably not getting those videos unless you look at third party rehosts.
- trinix912 1y agoWhat's even worse is they used to have great documentation, then for some nonsense reason archived it and replaced it with this wannabe JavaDocs crap with vague function explanations that leave more questions open than they resolve. They now tell people to watch WWDC videos as if those relatively short videos contained the same amount of information a proper API documentation does.
- spike021 1y agohonestly i don't see this as Apple-specific but more of a sign of the times. i've seen a few libraries and vendor platforms we use at work with the same style of documentation (really lack thereof).
- jiggawatts 1y agoMicrosoft's documentation quality very noticeably fell of a cliff about a decade ago. Now most of their public APIs and SDKs have hundreds of thousands of "documentation" pages that are just the name of the function with spaces added between the words. There might also be a listing of the parameters, which helpfully tells you that "string param1" is called "param1" and takes a string as the data type.
- spike021 1y agoYeah, the extreme basics seems to be the trend. We've actually had a lot of trouble due to this exact problem with one particular vendor now. Even repeated requests to improve their documentation are completely ignored or laughed off. it's too bad I don't have the purchasing power otherwise I'd suggest we should find a different vendor.
- whstl 1y agoI can attest to that. I used to freelance doing porting about 10 years ago, and the main way of having a stable event loop for a game on macOS is not really anywhere clear in the documentation, at least not anywhere I could find. Libraries and SDL, GLFW and Sokol handle this for you, and I had to poke inside them to actually figure out how to get the same performance of other games. In a nutshell, the trick is simple, you call [NSApplication finishLaunching] and do a while-loop yourself instead of calling [NSApplication run].