Closed Kerl13 closed 2 years ago
Good idea @Kerl13, thanks for submitting a patch! Other than the minor comment, LGTM.
Btw. I'm not sure how you intend to use boltzmann-brain
but you might be also interested in generic-boltzmann-brain
which instead of producing standalone modules (and by extension executables) lets you create multi-parametric Boltzmann samplers using template Haskell. I'd argue that it's much more convenient, especially if you intend to do something more with the generated objects.
See https://github.com/maciej-bendkowski/generic-boltzmann-brain
See https://github.com/maciej-bendkowski/generic-boltzmann-brain
I was doing performance comparisons against arbogen and usainboltz, so I just wanted an executable and boltzmann-brain
is perfect for that. I agree that the new tool looks better for most other use-cases ;)
LGTM, cheers!
As far as I understad, the current semantics of the --output flag is to expect an executable name (without any slash) and the generated executable would be copied to scripts/bin/ inside boltzmann brain's source tree
This seems counter intuitive to me. Instead I expected the following command:
to generate an executable named blah.exe in the directory /path/to. This patch implement this semantics.
What to you think of it?
By the way, I like the new interface of boltzmann brain better, congrats for that! It simplifies things a lot compared the old interface with the medulla middleware (although there are still some quirks I'll report in other issues)