hello! happy to be here

Hi

I have a somewhat vague question: is there a project related to modal logic and behavior driven testing?

Don’t know.

Searching for “modal” on http://pkgs.racket-lang.org\|pkgs.racket-lang.org gives no results.

Here are the result for “logic”:

https://pkgd.racket-lang.org/pkgn/search?q=logic&tags=

I think MiniKanren, Datalog and Parenlog are the the most “logic” related.

But I don’t know whether they are suitable for modal logic.

minikanren seems interesting

cannot find this datalog you mention

thanks

https://docs.racket-lang.org/datalog/datalog.html?q=modal

datalog may be useful for behavior testing

if you define some sort of test case database

The MiniKanren website has lots of useful resources:

http://minikanren.org/

Including links to papers and screen casts.

@orfeaskar What exactly is “behavior testing” and what makes it different from, say, the way rackunit tests?

https://docs.racket-lang.org/rackunit/api.html?q=rackunit

well it’s a current trend in system-level testing. Basically you have some workflows in your business that go through all kinds of systems

such as a customer requesting a package that has to be bundled together

and the end result is not in terms of just output of the workflow but also in terms of side-effects

you want to deliver the package but in a specific manner

https://hiptest.com/docs/about-2/

this is one implementation, in Ruby I think

thanks for the explanation

i hope it was helpful

for example, it may be that your package needs to be authorized by an identity server

a behavior test would be that the package will eventually be authorized if it is valid

but it will be necessarily the case so

i will try to do it myself, after all this is the point of racket - extensibility

@mflatt where do tests for struct/define-struct/struct-copy locate?

“pkgs/racket-test-core/tests/racket/struct.rktl”

Thanks! I was looking at pkgs/racket-test, so I couldn’t find it

@mflatt there’re struct-copy1 and struct-copy2. What do these mean? https://github.com/racket/racket/blob/master/pkgs/racket-test-core/tests/racket/struct.rktl#L1055

If I add a new one, should I name it struct-copy3?

A test in the core test suite is either (test <expected> <proc-expr> <arg-expr> ...) or (test <expected> <non-proc-label-expr> <expr>). So 'struct-copy1 and 'struct-copy2 are just labels for the test.

Right, but shouldn’t the labels be all different?

I see a lot of struct-copy2

The labels just need to be distinct enough that a failure can be racked back to a test.

So, probably the expected result or surrounding tests can distinguish 'struct-copy2 tests. Or it has been good enough so far, at least.

I see. Thanks

Sorry for a lot of questions. This should be the last one. Can I use (module xxx racket/base ...) in the test file (.rktl), and if not, what should I do instead?

Yes, you can write a module form. Pick a long name for the module so that it’s unlikely to conflict with any other module declaration in the tests.

I’ve got an unusual packaging question

I maintain a Racket package (herbie) and I’d like to have a plugin system of sorts for it

So that you could install another package (like softposit-rkt) and it could somehow notify the herbie package that it should load special support for it

@pavpanchekha You could probably use setup/getinfo to communicate information via info.rkt files. https://docs.racket-lang.org/raco/getinfo.html

Could you give a little more detail on how I’d set it up?

It’s how DrRacket’s plugin system works, for example: it looks for info.rkt keys named drracket-tools, drracket-tool-names, drracket-tool-icons, and drracket-tool-urls, as documented here. https://docs.racket-lang.org/tools/implementing-tools.html

Using setup/getinfo, you can choose your own key(s), like herbie-plugins or something like that, then call (find-relevant-directories 'herbie-plugins) to find info.rkt files that define that key.

So I’d use find-relevant-directories to search for a herbie-plugin key or whatever, then use get-info to read out the name of the module to load, and then dynamic-require it?

ha you beat me to it

Right, exactly.

Excellent, thank you so much

This is fantastic!

Good luck—I’ve never actually done anything like that, so I am mostly just guessing here. :)

Well, one of the most important things you told me is that DrRacket has a plugin system, and I can always read its source code to understand how it works

Yeah, that’s what I was mostly basing my suggestion off of, though I don’t know how complex the DrRacket code actually is that handles plugin loading.

Awesome, well, thanks so much for the help!

@pavpanchekha you might also look at how raco commands work, which is a little simpler than DrRacket but works basically the same way

Is there any way to ask Racket to make a gl-context<%> that uses OpenGL 3.3?

Ah, nevermind, it’s the set-legacy? method of gl-config%.