As of the changes in https://github.com/rmculpepper/crypto/pull/14 and https://github.com/Bogdanp/koyo/pull/28 you should no longer need to explicitly install libargon2-1 . If you point your catalogs to more recent snapshots, the .so should be included in your app’s distribution automatically.

(I realize you copied that bit from the congame repo and I’ll have to do the same there at some point)

Thanks and yes, the script is even hitting your Racket snapshot server. Would it be possible toremove some of those lines in raco pkg config? besides updating to the latest snapshots

I mean, I assumed you mantain http://racksnaps.defn.io/built-snapshots/\|racksnaps.defn.io/built-snapshots/. Maybe the packages needed are hosted too in http://download.racket-lang.org/releases/7.9/catalog\|download.racket-lang.org/releases/7.9/catalog and your server was added as a fail-safe?

I avoided GitHub actions for the time being. It would be a nice addition in the future but it is over my head now

And please excuse me if some of what I just wrote seems nonsensical, I just woke up and I am sipping my first coffee as I write :slightly_smiling_face:

Perhaps it would be also possible to include the npm build steps detailed in the README into the multi-stage build process of the Racket container. I’ll try that later

> Perhaps it would be also possible to include the npm build steps detailed in the README into the multi-stage build process of the Racket container. Yes, I typically have a separate stage to build assets and those then get copied into the distribution stage.

> Would it be possible toremove some of those lines in raco pkg config? All of them are optional, unless you want to use Racksnaps.

> Maybe the packages needed are hosted too in http://download.racket-lang.org/releases/7.9/catalog\|download.racket-lang.org/releases/7.9/catalog and your server was added as a fail-safe? That catalog is used for packages in the Racket standard distribution. It does not contain any user-created packages, so for user packages, Racksnaps is used. <https://defn.io/2020/05/03/ann-racksnaps/|This post> has some more details.

If that’s a bug, contract-name may have a similar bug. I recall correctly, for years I’ve had to pass things through coerce-contract before contract-name or I’d get an error. Even though something like boolean? or 12 might count as a contract?, it apparently isn’t a contract in the way that matters for contract-name.

rename-contract seems to work on numbers, now

Oh, I might be recalling incorrectly.

Hmm, on 7.5 I’m seeing contract-name works well on boolean? and 12. At some point, I might have to remove the coerce-contract calls in my code and see what breaks.

when calling a foreign function via the ffi that reads/writes on std in/out is it possible to something akin to call-with-input-bytes and call-with-output-bytes around the call?