Rewrite is coming

[David Pineau]

Greetings people !

As you may all be aware, it's been about one year since we stopped
working on any new feature in the language and the compiler, and stuck
to bugfixing (or at least trying to) the current compiler.

During this Week-End's FOSDEM, we got the opportunity to talk about
the state of the python tools for the rewrite to come.

Sooo, to sum up the state:
- The Pyrser tool is functionnal and can be seen as finished.
- The cnorm library (V4, in pyrser) is also quite complete and functionnal.
Both have been used by students at Epitech; so it's already a good point.

The only lacking thing in pyrser is currently a typing module that
lionel is currently working on.

So here is what we're going to do:
Lionel and me are going to work together on the typing module, which
will be used extensively throughout rathaxes. This is the perfect
opportunity for me to embark and learn python(3) while working one one
of the core modules of the new compiler.

Also, since we are starting anew, we decided to resume the old habit:
1 hackathon per month. For the sake of simplicity, we'll choose the
first full Week-End of every month for this hackathon, instead of
organizing each meetup one by one. In case of impossibility, we may
move the hackathon to another week end, but that should stay as rare
as possible.

For the new guys joining us on the compiler route, I strongly advise
you to try and read what Lionel and me are going to commit, and stay
up-to-date with the compiler.

Louis will still be working on the drivers, but will try to be much
more active on the compiler side, in order to stay up-to-date and be
able to debug things himself, and rely less on my humble self !
In the end, we are still trying to find a good way for us to stay in
sync, despite our different timezones and the distance. If you have
any idea, you're welcome ! Otherwise, we currently plan to have
little meetings outside the hackathons to do a point and review the
code updates together.



Although it seems a bit optimistic, I'd actually love to reach the
current state of the compiler by the next FOSDEM, and that should
explain a bit why we are trying to get back on the road in this way.


Thanks for reading me !

--
David Pineau

[Louis Opter]

In addition to what David said, I'm attaching the short assessment we
wrote and discussed during the Fosdem about this second iteration of
the compiler (the first iteration being Rathaxes 2009).

It basically summarizes what we learned in the past two/three years on
the compiler, and highlight some errors and designs issues that we
must keep in mind in this third iteration of the compiler.

One of the biggest mistake we made was to over-engineer some aspects of
the language prematurely and it turned out that what we actually needed
to move forward was not working. To avoid repeating this mistake we need
to make the feedback loop between the drivers (i.e: the language) and
the compiler as tight as possible.

With this document, I'm going to push my current state of work on the
e1000 driver (from a feature point of view it's a regression compared to
what we have, but from a language/architecture point of view it's much
more correct).

We are also going to tag the current version of the repo and replace
some parts of the tree with new code soon.

Brace yourself.

--
Louis Opter