316 karma · joined March 30, 2016
Once you're into testing I/O, which is ultimately unavoidable no matter how hard you try not to, you either need cooperative third parties who can give you truly representative test systems (rare) or a certain amount of test-in-prod.
Testing database stuff remains hard. You either wrap things in a some kind of layer you can mock out, or dupe prod or some subset of it into a staging environment with a daily snapshot or similar and hope any differences (scale, normally) aren't too bad.
Copy-on-write systems or those with time-travel and/or immutability help immensely with test-in-prod, especially if you can effectively branch your data. If it's your own systems you are testing against, things like lakefs.io look pretty useful in this regard.
And yes, feature flags, good metrics, and load balancers that you can send a small percentage of traffic through a new version (if your traffic/system allows such things) all help.
I then hired a dude with a LIDAR scanner and did it properly. The difference in quality/accuracy is like 120x80 ASF video files in 1996 compared to 4K footage today.
Anyone who thinks you can build "HD" virtual worlds using the crappy cameras on a Tesla needs their heads examining. Maybe with thousands of passes and some epic compute and signal processing, but why bother? Just LIDAR it.
My Tesla can't even decide if a traffic light is a single traffic light or not on a sunny summer's day from a distance of twenty feet. Almost every time it is either dark or humid or winter (road grime) it tells me one or more cameras are obscured. But only after I've already started driving, obviously. This supposedly cutting edge AI driving machine frequently thinks I'm leaving the carriageway on UK B-roads (it's almost dangerous) and is significantly less reliable at distance cruise control and lane-assist than my Skodă. (I presume VW just quietly bought a black box from Bosch or whoever to do this.)
Tesla are barking up the wrong tree IMO. At this point the camera-only stance feels like a religious thing, not based on sanity or the real practical world. I imagine that someone came to Elon and said "reconciling conflicting radar and camera signals is hard" and he applied his considerable genius and issued an edict to "let's not do that then!" like it would magically make all the actual hard problems go away.
Heck, Teslas can't even seem to reliably parallel park themselves, frequently getting stuck halfway, or hitting kerbs. If they can't solve that highly constrained problem, I'm hardly going to trust taking my eyes off it at 70mph.
You might be able to build your perfect todo app with it.
Don't like workflows? Don't add a workflow field.
Just you and no team? Don't add an assignees field, etc.
Want unlimited nesting? Set up a relationship that lets you do that.
Need recurring tasks? Set up an automation that reopens or clones tasks with a recurring field you've added, every week or day or whatever.
Want to know which tasks are "available"? Set up a formula field for that which looks at start date, the task statuses of any blockers, and whatever else you like (time of day even).
It's very cool.
They only supported the 5C for two years after they sold the last device.
That said, NUnit has much better syntax for this, where you put parallel/multiple assertions like this in an explicit block together: https://docs.nunit.org/articles/nunit/writing-tests/assertio...
Much less boilerplate duplication in the actual test framework, too.
"We can't land this tiny fix because we have a release planned within three months" sort of thing.
How do you think that debt ends up getting paid off? It'll be close to $1 billion every year just to service that debt.
Given that:
* Twitter has only ever made a profit in 2018 and 2019
* That the profits it made in those two years were $1.2bn and $1.4bn respectively (only just enough to service this new debt)
* That for 2020 and 2021 other tech companies were comparatively minting it, but Twitter burned ~$1.3bn
* The world is generally in the early phases of what is likely to a deep recession, with ad spend and consumer spending already well down
It feels like a bit of an interesting purchase, to say the least.
It should be noted that the plan to charge users $8 per month for a blue verification badge wouldn't even service the debt if 25 times more people bought it than currently have it for free.
The only thing it's really lacking is granular (per-field/entity) access controls, which is on their roadmap (although it's worth noting it's been on their roadmap for quite a while).
"Improve functionality over time" implies that the basic functionality of full self-driving is there already, and that it will be improved. No reasonable person would take it to mean "does not exist yet, and may never do".
They would also take "capable of" to mean "right now", not at some nebulous point in the future, especially given the context of the feature being a paid upgrade.
The definition of "full" is "not lacking or omitting anything; complete". One might reasonably expect "full self-driving" to mean a human does not need to intervene or monitor. I.e. What the industry commonly calls autonomous driving level 5. They clarify what they mean elsewhere on their web site:
> Autopilot, Enhanced Autopilot and Full Self-Driving Capability > are intended for use with a fully attentive driver, who has their > hands on the wheel and is prepared to take over at any moment. > While these features are designed to become more capable over time, > the currently enabled features do not make the vehicle autonomous.
So they don't even claim level 1 autonomy.
Even with your version, the hardware may very well never be capable of full self-driving at all. There's no evidence to suggest it will be - hopes, dreams and promises don't count. And they've done an upgrade on the earlier-gen Model S cars already, where they presumably made similar claims.
The whole thing is misleading at best, downright trading standards fraud at worst.
For what it's worth, my Model 3's carriageway departure warning kicks in randomly and the amount of phantom braking it does even on basic cruise control is terrible. Autopilot is worse. I find monitoring it in autopilot mode significantly more stressful than just driving the car myself. I wouldn't trust FSD at all, and certainly not enough to want to fork out the silly money it is.
I.e. Lightning-fast comes before Powerful.
If it fits neatly into the built-in data structures or smells quite systemsy and C-like, it's probably fairly easy in Go.
String and array manipulation are similarly painful to C. The lack of generics can be a massive pain in the neck, meaning you end up copy/pasting things, and other people's code doesn't tend to be very easily extensible without actually editing it to add the appropriate interfaces or whatever. The lack of OO is sometimes fine, sometimes not. It's really not a very expressive language. (I appreciate that's a rather fluffy/vague thing to say, but I feel it's true.)
On the other hand, you can cross-compile stuff trivially, you get nice self-contained binaries, it's generally sufficiently performant, and it's simple in the same kind of way C is, only with a garbage collector and easier threading (goroutines are pretty nice).
There's no more ceremony really compared to something like Python+Flask, but the business logic you end up running behind your web API calls _may_ be a bit more verbose and feel clunkier. Or if you're lucky it may not. It's very much dependent on what exactly you're doing.
I get that you could in principle namespace things (at least for package managers that support this) and insist on a small set of company-internal signing keys for those namespaces. But managing all that isn't easy and what about for package ecosystems that don't really have namespaces (e.g. PyPI, NuGet)?
In e-mail terms: - Element is like a mail client - matrix.org is like a given community of mailing lists
In IRC terms: - Element is like an IRC client - matrix.org is like an IRC server/network e.g. Freenode.