3 ms·
I'd have to register my vote on this not clearing things up. All it did was talk about relatively simple sections of code with well understood boundaries. Ins
by taeric 9y ago
I'd have to register my vote on this not clearing things up. All it did was talk about relatively simple sections of code with well understood boundaries.
Instead, I often find people that try to make "units" of code not "Money" or "Transaction" but crap like "WellsFargoTransaction" or "EuroToUSDollarTransaction" complete with separate classes for each of the things that those imply.
And the amusing dig against mutable state is... well, unnecessary? It didn't even really do anything other than signal to a certain crowd that he is on their side.
Integration tests may be a symptom of poor design. I don't see it, though. Integration tests should exist to test the assumptions that your unit tests rely on. Because, by necessity, your unit tests will be making assumptions about the rest of the systems they interact with. To claim that some "pure" design can get you away from needing to test those assumptions is interesting debate starting rhetoric, but a dangerous goal to put in people's minds.
- AstralStorm 9y agoAbout the only such pure design would then be fully formally verified and specified, also across module boundaries. Good luck with that.
- taeric 9y agoCompletely agreed. It is funny, because I do think this is why we are constantly reimplementing things in the new languages, instead of finding ways to connect old programs. That is, people are so worried about bugs in hooking grep up to their system, that they instead reimplement all the bugs building grep into their system. (Where grep is an easy to type example from my phone. :) )