I find I am regularly running roc test
and then roc dev
... given the distinction between check, run, dev
would it be a good idea to have roc dev
also run roc test
? I would also ask for roc format
to be included, but it's not quite at the same level of maturity. OTOH that might be beneficial for finding issues bugs?
I am continuously using expectations to check my logic as I go along. Though, I have been starting to use roc repl
more for testing smaller bits of code, however it doesn't support expect
calls yet.
I'm also interested to know what workflow people are using? I assume it's difficult to just write, compile and it just works. I'm looking forward to experimenting with these ideas in the editor
.
would it be a good idea to have roc dev also run roc test?
I'd say we definitely want the ability to run roc dev
without roc test
.
Why is that? I assume it significantly slows down the compile loop, but I also imagine you could use roc run
if you just wanted it to run fast. What about something like roc dev --test
or roc dev --no-test
?
@Luke Boswell I moved the previous messages to their own topic.
For context: dev
compiles fast so you can iterate quickly for development. run
compiles slower to create the fastest possible binary.
Since dev
is all about a fast feedback loop I don't think we should add things that make it slower :p
What about something like roc dev --test or roc dev --no-test?
I would expect people to use roc test && roc dev
. Providing redundant ways to do the same thing does not seem necessary to me, but I welcome people to convince me otherwise :)
Yeah exactly, it's a core design goal to have some command that just does the fastest possible compilation and that's it. And tests take time to run. You often want tests too, but not absolutely always so we shouldn't force that.
I think tests fit much better with some sort of watch command to auto run on update of files. Definitely requires limited scope/caching as projects get bigger.
Last updated: Jul 06 2025 at 12:14 UTC