Project roadmap and goals

[Jon]

As a 75% Windows, 25% Ubuntu-based-distro user who's just started looking at Rubinius and would like to better understand your current thoughts and goals regarding Windows support as I've not really found anything recent in the docs or searching this group.

Many times this question garners tangential or politically correct, on-the-surface responses so I'll try to be very clear as to why I'm asking the "Windows support" question.

* I fully understand and appreciate the need to focus your resources.
* I understand that many things will change.
* I am not looking to defocus anyone, nor nail anyone down on support commitments, nor request roadmap changes.
* I understand I'm on my own for toying with Rubinius on Windows.

Really I'm just looking for insight as to what you guys are currently thinking and bit of expectation setting.


1) Do you ever see running Rubinius on Windows becoming a priority?

2) If yes to (1), what would need to happen for it to become a priority?

3) What specifically, if anything, would you currently like to see from the community regarding Rubinius on Windows?

4) What areas of the current codebase and build system do you think are problematic for Windows?


Regards, Jon

[Evan Phoenix]

Hi Jon,

On Nov 5, 2009, at 9:49 AM, Jon wrote:

>
> As a 75% Windows, 25% Ubuntu-based-distro user who's just started
> looking at Rubinius and would like to better understand your current
> thoughts and goals regarding Windows support as I've not really
> found anything recent in the docs or searching this group.
>
> Many times this question garners tangential or politically correct,
> on-the-surface responses so I'll try to be very clear as to why I'm
> asking the "Windows support" question.
>
> * I fully understand and appreciate the need to focus your resources.
> * I understand that many things will change.
> * I am not looking to defocus anyone, nor nail anyone down on
> support commitments, nor request roadmap changes.
> * I understand I'm on my own for toying with Rubinius on Windows.
>
> Really I'm just looking for insight as to what you guys are
> currently thinking and bit of expectation setting.
>
>
> 1) Do you ever see running Rubinius on Windows becoming a priority?

Absolutely! Largely we've been waiting for someone interested in
Windows to come along and help us out.

>
> 2) If yes to (1), what would need to happen for it to become a
> priority?

Someone interested in Windows and Rubinius to show up! If that didn't
happen (which would be sad), we'd prioritize it after we feel like
unix support is mostly done.

>
> 3) What specifically, if anything, would you currently like to see
> from the community regarding Rubinius on Windows?

Atm, it would largely be someone with knowledge of porting unix code
to Windows. We don't use a lot of unix only features, but a number of
them (pthread, etc). We try and abstract those kinds of things as much
as possible, but I'm sure there are holes in the abstractions.

Someone to go through and extend or add abstractions for Windows is
the primary thing.

>
> 4) What areas of the current codebase and build system do you think
> are problematic for Windows?

This is probably the most unix centric part atm, because we expect g++
a few explicit places. That being said, we use rake for the primary
building, so that should run fine under MRI on Windows. My guess is
that building with mingw would be the easiest to get up and running.

Thanks for the interested. If there is anything we can do to help you
along with Rubinius on Windows, just let us know!

>
>
> Regards, Jon
>
> >
>

[Jon]

> > 3) What specifically, if anything, would you currently like to
see  
> > from the community regarding Rubinius on Windows?
>
> Atm, it would largely be someone with knowledge of porting unix code  
> to Windows. We don't use a lot of unix only features, but a number of  
> them (pthread, etc). We try and abstract those kinds of things as much  
> as possible, but I'm sure there are holes in the abstractions.
>
> Someone to go through and extend or add abstractions for Windows is  
> the primary thing.

This answers a few followup questions. Thanks.

> > 4) What areas of the current codebase and build system do you think  
> > are problematic for Windows?
>
> This is probably the most unix centric part atm, because we expect g++  
> a few explicit places. That being said, we use rake for the primary  
> building, so that should run fine under MRI on Windows. My guess is  
> that building with mingw would be the easiest to get up and running.

Understand and I primarily use MinGW or http://www.tdragon.net/recentgcc/
although the MSFT compiler story is better now with the Express
editions and the command line compilers included in the Windows SDK.

FWIW, I'm looking the following cross-platform (binaries available for
Windows, Linux, and Mac OS X (Universal)) Lua-based build tool and you
may want to check it out if you aren't aware of it already. I like
what I see so far as it's just a ~230K executable that's needed to
use...no other dependencies other than a C rt I believe.

http://industriousone.com/premake

..and an example build config script

http://code.google.com/p/ontl/source/browse/trunk/samples/psinfo2/premake4.lua?spec=svn640&r=640

I'm also hearing interesting things about http://code.google.com/p/waf/