R13 Measuring DX

Why should there be a different framework and application design for single user, vs multithreaded, vs cluster vs web scale ? Why are backup patterns, recovery, availability all different at these different levels ?

how would you measure the DX of a particular framework ? What makes it good or bad compared to another one ? Is the best we can do to conduct a survey ? rexmondo — Today at 10:49 AM There’s some simple standard measures like cyclomatic complexity Also think about it like ergonomics You wouldn’t sit in a chair that put huge sideways pressure in your knees Using a dev framework that forces you to keep 1000 things in mind every time you’re writing a line is like putting strain on your brain I suspect it’s possible to build tests that can discriminate high mental load frameworks from low ones Kind of roughly counting how many dependencies, layers of contextual surrounding code, special rules exist in your average code unit mcnasty — Today at 11:42 AM so if I request "developer experience" as a requirement for Interblock, how would I know someone had met that ? A simple way could be to select a collection of other stacks that can be used to build an application, and just put them in order of hard to easy ? so it becomes a relative measurement ? if someone runs some metrics on a chair, I might not know what that means, but if I can sit my ass in them, I can definitely rank them in order of preference there was this programming site I saw a while back that implemented a bunch of simple apps, like todo list, using different dev frameworks, and used that to derive metrics for ? can't damn well find it tho... rexmondo — Today at 12:46 PM Yeah todo list is a great example of how devs sniff this stuff out There’s a standard ui for it but it’s slightly different in each framework It’s going to be a bit of a tougher sell with inter lock because most of the todo list implementations are client only I think to do a real apples to apples DX consideration you’d need to show differences in app lifecycle management including migration, scaling, refactoring mcnasty — Today at 12:48 PM maybe theres a todolist style app we can do that is for a distributed app as a starting point, how about a truly decentralized todo list - that starts the test at basically impossible for conventional frameworks rexmondo — Today at 12:49 PM Yes That’s a great idea mcnasty — Today at 12:50 PM then we can make a bunch of metrics statements about it, and make some equivalents in other blockchains as a good metric, wasted data % is a good one - as a client of this app, how much data am I receiving that has nothing to do with me ? rexmondo — Today at 12:50 PM I think even if it was possible in conventional frameworks but the amount of code is staggering mcnasty — Today at 12:51 PM for traditional blockchains this will be the majority of data then we can define some contrived lifecycle events: make the todo app be multi user, upgrade one dependency library, upgrade the framework version without bugs, scale to 10k users, simulate being slashdotted at each stage, include costs for infrastructure, and also for dev time rexmondo — Today at 12:53 PM Yes I love this Like a case study and a tutorial in one mcnasty — Today at 12:55 PM then torture test: netsplit, slashdotted, patch the framework, and release 1 new feature while users are offline 🤣 like that's not that contrived, but you're testing so much more than just a code framework there - thats software supply chain integrity and availability rexmondo — Today at 12:58 PM Yeah that’s the ticket Instead of having those all be unaddressed future concerns that everyone answers differently, you have the answer up front I think fear of real world conditions holds a lot of people back from trying weird stuff in dev And for those of us it doesn’t, we get stung by things we didn’t see coming mcnasty — Today at 1:00 PM yeah totally I think the final test is how easily can you get your code written, and how much will it cost you, and how much will it cost to maintain those are the metrics of the dreamcatcher if I can take 15 minutes to drop in a spec for my todo list, $100, and review a few responses, thats the best framework of all plus if it automatically has people fixing bugs for me, and auto upgrading the base library after ensuring that my code executed exactly the same as it always did #winner rexmondo — Today at 1:03 PM Yeah man mcnasty — Today at 1:04 PM oh yeah and do all this offline too ! 😆 distributed offline mic drop 🎙️ rexmondo — Today at 1:04 PM Yeah that’s a good point mcnasty — Today at 1:05 PM if a mic drops on the internet, but no-one is connected to hear it..... ok well this sounds like a great case study, tutorial, spec / goal thing - I'll write it up