💾 Archived View for heavysquare.com › notes › FFEB-it-hard-to-test-everything.txt captured on 2021-12-03 at 14:04:38.

View Raw

More Information

-=-=-=-=-=-=-


~~ 2021-02-02T16:35:41+01:00

I have the impression that the space of possible test cases for a given
piece of software is exponential in order. Like you have a really simple
thing and you could do A and B to it in a sequence and have expectations
of its behavior.

Is it starting up ok?
After a single A, ok?
Single B?
What if I do A and then B?
Opposite order?
What if I do two As and then a B? Still ok?
etc

[AB]* is having loads of elements

Let's say this software supports batched execution of events, like
sending AAAB in a batch.

What if I send two As in separate batches?
What about an empty batch at the end?
One batch of AA and an empty?
etc

Eg here in the second case, it'd be much more convenient to have
guarantees from the program code, or language constructs, so that batching
is definitely not altering semantics.