Antony Courtney

I cannot possibly stress enough how this is the reaction you get from literally everyone who writes code for a living. Use computers to speed up unofficial counts. Make sure there's a human-verifiable paper trail.https://twitter.com/ASpittel/status/1224544431426109440 …

Once you know the API you're designing is good, the UX you've created is good, and the UI design has settled, Sure, write some tests. But they're not helpful before that. Also, with great dev tools tests are pretty useless generally (outside of product specific algorithms) IME.

What stage of the game is your product and how much money do you have? Shipping a complete (though buggy) product is WAY better than building 30% of a fully-tested program that you can't ship. For most products in the beginning, very few tests have any value at all.

Also, you have to realize what kind of program you're building. Are you building a pillow fort that will be replaced the next time the designers look at it? Or are you building a castle that will be continually improved and upgraded? Not all code needs tests.

Like what if public agencies had the staff and the mandate and the resources to do their jobs and provide good services to everyone, and nobody needed to call in favors because stuff was handled the way it's supposed to be?

The viral thread quoted below is missing essential context and contains numerous errors. It does not reflect the latest evidence. #2019nCoV Here is a new thread with the facts: https://twitter.com/DrEricDing/status/1220919589623803905 …

That's the secret. Do a shitty job. Then, if you need to do a better job, start from the shitty job and make it better. If you can't do a better job, ask for help with the shitty job as a starting place. If you can't get help, well, turn in the shitty thing.

Front end devs: I'm here because I like programming *and* design. Back end devs: You're not a real programmer. UXers: You're not a real designer.

For big projects, especially open source ones, modularity isn't just good engineering practice. It significantly increases the chance of success of a project, because even if the overall project isn't popular subcomponents can be.

TODO doesn’t mean you actually intend to do something. It’s a marker for the next person that the logic was left unfinished and they shouldn’t be surprised by missing corner cases or strange code structure.

I think the biggest GC-related performance benefit of value types isn’t that you can allocate them on the stack, since a generational GC will easily eat up those allocations. It’s so that more *long-lived* data can become pointer-free and not scanned at all.

Just ordered a couple 1TB thumb drives for $30 each. I still find myself awestruck by tech progress -- I desperately wanted a 10MB hard drive for my IIGS as a teen, but didn't have the $399. One million times cheaper per byte, one thousand times faster, and 100 times smaller now.