Re: Emacs rewrite in a maintainable language

[John Wiegley]

>>>>> Óscar Fuentes <address@hidden> writes:

> Many times, while looking at the Emacs C sources, I thought "this would be
> much simpler to understand and modify with C++." Furthermore, I would
> volunteer to work on it.

Please, not C++. I don't believe the result would be any easier to maintain
overall. The difficulties would just be shifted. Further, our candidate
contributor pool would shrink.

One of my open source projects, Ledger, is in C++. Despite having many active
users, the number of pull requests I've received for the C++ code can be
counted on both hands. Meanwhile, pull requests for ledger-mode continue to
come in, and I was able to find a full maintainer for that part of the code.

The thought of an Elisp->C compiler that could produce well-performing code is
very interesting. I starting working on an LLVM JIT for Emacs bytecode at one
point -- got as far as hello world -- and it was not terribly difficult. The
main reason I stopped is that it didn't help performance enough. The slowness
imposed by constantly looking up symbol names at global scope, due to dynamic
binding, dominated the profiling results.

A subset of Elisp that doesn't have dynamic binding might well produce code
that could be of use to us, and wouldn't need any of the macros we currently
use to communicate between the two languages.

John