[neomutt-devel] mutt vs neomutt

[Guyzmo]

On Mon, Feb 13, 2017 at 11:13:40AM -0800, David Champion wrote:
> What platforms lose support?  I think it's OK to decide on a case by
> case basis, but here's what I expect:
> * anything no longer supported by the manufacturer
> * anything we can't get anyone to test on
> * OSes older than X for which there are updates (X is variable)
> * OSes with irremedial, significant security issues
> 
> I would NOT extend this to making explicit choices to support only
> certain compilers -- that puts our support choices into other people's
> hands.  But I do think that, similarly, it's OK to decide ad hoc that a
> valuable patch that breaks support for Sun Compiler Suite for Solaris
> 2.5 is okay.  You can get gcc for that.

there's been a discussion about that at some point either on this ML or
on an issue, and as I really think there needs to be regression tests
and CI for mutt, we need to have a "tier 1"/"tier 2" support scheme:

- tier 1: everything we *actively* run tests on (Linux, .*BSD, Darwin, Mingw…)
- tier 2: everything we that we know we can compile on and do our best to not break

"Tier 1" would mean having a stamp "tested with compiler X on system Y",
and however you like it, we're likely to test only with clang and gcc,
which are working almost everywhere.

Also, I have been suggesting to rebase all the I/Os on libuv, which
would add asynchronous I/O (and network) handling, using a well and
thoroughly tested library (used by nodejs, neovim and many other
projects).

The downside of such a choice would be to have strong dependency, and
kill support for some exotic systems (not sure what those would be, I'm
not even sure the libuv team knows).

Cheers,

-- 
Guyzmo