16:32 <jemc> wow, just had the realization that incremental compilation will be a lot easier to implement in the pony-compiler-in-pony than I first thought - thanks to the `serialise` package, we can get a nice compact serialization of any internal data structure "for free"

16:46 <staticassert> what's the state of the pony pony compiler?

16:51 <Xyliton> recover can turn a tag into a iso, right?

16:53 <Xyliton> this is my current code: https://gist.github.com/TheFreakLord/9e1b0a782e188603dcc08855089c98c1

16:53 <Xyliton> it complains that tag is not a subcap of type iso, on line 79

17:54 <SeanTAllen> xyliton: you can not turn a tag into an iso.

18:00 <jemc> staticassert: it's in the "vapourware" stage ;)

18:01 <jemc> though working on it is currently my main priority (in addition to working on other things I keep finding to be blocking the effort)

18:02 <jemc> but the more I work on the compiler, the more I keep running into code that could be cleaner, or issues that could be completely avoided if we were using Pony as the implementation language

18:04 <staticassert> yeah, I think I'd be much happier to work in pony than C

18:05 <jemc> the incremental compilation thing just happens to be a big example of that - I've been sizing up what it might look like to implement incremental compilation in the current C-based pony compiler, but serialising the internal data structures to a file in an efficient way would be a pretty significant challeng in that effort, and one that could be completely avoided in the pony implementation

18:23 <staticassert> that's pretty cool - I'm unaware of how incremental compilation really works

18:23 <staticassert> I didn't realize serializing was even necessary but I guess it would have to be

18:59 <obadz> jemc: in-editor typechecking/hints would also be lovely and I doubt this would happen before self-hosting

18:59 <obadz> (https://github.com/Microsoft/language-server-protocol ⇐ this looks like the right way to cover all text editors these days)

19:40 <jemc> agreed