I really don't like this code emission thing you made. I tried very hard to emit AST properly, connect is supposed to return an expression. This is not enforced by the type system (because we don't use metaocaml yet) but it is respected in the whole code base.

If functoria provides a nice way to do this, feel free to change my code. However, the version in master currently generates code that is broken (does not compile) and would initialise in the wrong order even if it did.

Will merge this today unless someone suggests a better solution, on the grounds that working code is better than elegant code.

Yes, there is a way, by modifying the prelude, but I don't really have time to try it.

Perhaps we should make an issue with a few notes on the problem so we don't
forget? We can then ask Amir to bring it up occasionally at the weekly
meeting for extra pressure ;-)

On Sun, Nov 29, 2015 at 4:46 PM, Gabriel Radanne notifications@github.com
wrote:

Yes, there is a way, by modifying the prelude, but I don't really have
time to try it.

—
Reply to this email directly or view it on GitHub
#466 (comment).

Dave Scott

Yes-- @Drup if you could give some indication of how you'd go about it,
perhaps someone else will take it on.
Ultimately I don't want to hold up the functoria merge because it breaks
profiling-- and although, as @talex5 says, working code beats elegant code,
it would be nice to try and keep this more elegant than not!

On 29 November 2015 at 17:51, Dave Scott notifications@github.com wrote:

Perhaps we should make an issue with a few notes on the problem so we don't
forget? We can then ask Amir to bring it up occasionally at the weekly
meeting for extra pressure ;-)

On Sun, Nov 29, 2015 at 4:46 PM, Gabriel Radanne <notifications@github.com

wrote:

Yes, there is a way, by modifying the prelude, but I don't really have
time to try it.

—
Reply to this email directly or view it on GitHub
#466 (comment).

Dave Scott

—
Reply to this email directly or view it on GitHub
#466 (comment).

Richard Mortier
mort@cantab.net

That's the thing, I didn't tried, so I don't really know. The base idea is to make the prelude parametric, so it will be different if needs be.

The issue here is not that it's inelegant: it breaks functoria's assumption about code emission. We can get away with it now, but I can't guarantee that it will stay that way.

(But sure, let's not delay release for that)

Merged. I'll create an issue with the notes from this thread to mark this as violating functoria's assumptions.