OCaml hacking session at MIT, Cambridge *US*, on June 6th

I hope you all have an excellent hacking evening today! Take some pictures :slight_smile:

1 Like

These tasks about the OCaml website may also be a good starting point — some of them do not require much expertise.

3 Likes

Really psyched to discover that this is happening.

Relative newcomer here - I’ve been doing hobby projects in OCaml for <1 year, and am looking forward to the opportunity to come and hack! Planning to bring an OCaml-inclined friend as well.

4 Likes

(At the last minute, of course.)

I created an organization wiki page to just add stuff for everyone to access, and a log wiki page for everyone to record their participation.

If you participate to the hackaton (physically or just online or in some other physical place), please add yourself to the log wiki page!

2 Likes

Thanks @gasche! Looks like it went really well :smiley:

From your log, I was interested to see that @bthom tested the RISC-V backend – I was considering setting up a qemu-based CI for it for OPAM in anticipation of getting some real hardware soon, so keen to hear how the testing went!

Yep, it was funk, thanks!

(I still have some cleanup work to do; I plan to transfer some of the “temporary page” organization to the normal wiki, and maybe restructure the log.)

So it turns out that bthom wanted a cross-compiler rather than a compiler, and I think the previous backend was written as a compiler (intended to be built itself within qemu). Most of the effort went into him discovering how to hack the ./configure for cross-compilation (we looked at how @whitequark does his cross-compilation trees but they directly contain the post-configuration output, and you still need to first play with ./configure to figure out what those outputs should do). Finally, he wanted to run bare-metal RISC-V program, so last I know he was gutting out the parts of the (target) runtime depending on dirent.h and other OS things. I don’t think that much actual testing was done (the one part of ./configure that actually execute binaries was run under spike, if that counts), but hopefully that could happen in the future.

I fleshed out the log page a bit more, with the intent of making sure that its content will keep making sense in the future (it can be archived somewhere or stay where it is).

I’ll work on moving the event-specific list of tasks into the general wiki pages. Then I think the “temporary page” used for coordination/organization can be archived or even deleted.

2 Likes

Since there is interest in centralising the compiler hacking activities now that they are going more global, there is also an issue to discuss the move here: https://github.com/ocamllabs/compiler-hacking/issues/12

Feel free to chime in with your opinions, especially if you want to host your own session!