> On Jan 18, 2020, at 3:48 PM, Gregory Nutt <spudan...@gmail.com> wrote:
>
>
>> This is not to say that this is a 100% stable development platform. There
>> are still a few dangling, loose ends that I am aware and most likely more
>> that I am not aware of. ... So basic functionality is there, but any real
>> use would need to spend a few days in further clean-up. I think with a
>> little day-by-day usage it could mature the rest of the way quickly.
>>
> In retrospect, that is an overstatement. Working with this more, I can see
> that there are several loose ends that would need to be cleaned up. It seems
> to build better than it actually is and the complete clean-up would probably
> require more than "a few days."
>
> The DISCUSS phase ends tonight but I will wait until Monday to start the vote.
>
I believe Windows Native support is always going to be maintenance headache and
should be deprecated; and flagged for eventual [immediate?] removal. My
thinking is:
1. we already build on Windows 10 with WSL installed (Windows Subsystem for
Linux) where we have a familiar POSIX environment and [mostly] compliant shell
(BASH). This should be the recommend / preferred way to build NuttX on a
Windows 10 host.
2. to support Windows versions pre-WSL (Windows 7, etc.) we can improve the
documentation on how to install Cygwin bash, make, etc.; and again rely on our
POSIX build system. We can even include a script to test the environment for
all necessary user land utils.
3. with the above two options, I don’t think we’re leaving anyone out. And we
end up with a cleaner, much more maintainable build and test gestalt.
comments/flames welcome…
