GHC on Windows (extended/broad discussion)

Simon Peyton Jones simonpj at microsoft.com
Tue Oct 28 22:02:26 UTC 2014


The people problem is tricky. At work, this would be the right time to do a video chat and at least see the faces of the other people involved. Would folks be interested in a Skype/Hangout sometime? It would be interesting to hear what interests / skills / resources / constraints we have between us.

I think that’s a great idea, thanks.  It’s easier to work with people with whom you have formed a personal relationship, and a video conf is a good way to do that.

Simon

From: ghc-devs [mailto:ghc-devs-bounces at haskell.org] On Behalf Of Gintautas Miliauskas
Sent: 28 October 2014 16:08
To: Austin Seipp
Cc: kyra; ghc-devs at haskell.org
Subject: Re: GHC on Windows (extended/broad discussion)

Hey Austin,

thanks for pushing this forward. It sure looks like Windows deserves more attention than it is getting.

We definitely need a broader action plan. My thoughts were close to yours:

1. Push through the gcc compiler upgrade (D339)
2. Fix the Windows continuous builds. This is necessary to prevent regressions.
3. Make sure validate.sh results are clean on Windows. Tests that are known to be failing are not providing new information, they should be disabled and issues filed.
4. Triage the Windows bug list. I already made a few passes, but most of the bugs are far from trivial. I think we'll need to prioritize very aggressively to focus the limited resources.

Are there any broader ideas for architecture-level changes related to GHC on Windows?

The people problem is tricky. At work, this would be the right time to do a video chat and at least see the faces of the other people involved. Would folks be interested in a Skype/Hangout sometime? It would be interesting to hear what interests / skills / resources / constraints we have between us.


On Sat, Oct 25, 2014 at 2:24 AM, Austin Seipp <austin at well-typed.com<mailto:austin at well-typed.com>> wrote:
Gintautas, Tamar, Roman,

(CC'ing those on
https://ghc.haskell.org/trac/ghc/wiki/WindowsTaskForce, and Kyrill,
who has helped us out much in the past)

Thank you all for all your help with Windows recently. I apologize for
not responding to some of your concerns sooner in the recent threads
about tarballs, etc.

First off, all your contributions are extremely welcome - GHC has had
many talented Windows hackers in days long past, but these days this
number has dwindled! Anyone who has an interest in GHC on Windows is
in a place to make a big impact and help us. All the work Gintautas
has done for example, will dramatically improve the ghc-tarballs
scenario.

On that note: Gintautas, I will get D339 merged in ASAP, as soon as I
test it and make a download mirror for you. Haskell.org has an awesome
new CDN setup, and once I implement https://downloads.haskell.org, it
will be easy to update tarballs and serve them to mass amounts of
users.

However, beyond that, we still need more done. First off, if you can
help, we can help you! We can make lots of Windows build bots for
people on demand, so if you're in desperate need of disk space or your
computers are a bit slow, we can help accommodate.

Right now, we have nightly builds with Gabor's[1] build system, and
soon, we're working on a Phabricator integration, which should be
great - and hopefully reduce the amount of breakage substantially.

I also notice there is a ticket list of Windows issues[2], and that's
fantastic. After a quick glance, a lot of these tickets are old,
duplicates, or could possibly be closed or fixed easily. A good first
task for any new contributor would be to go through this list, and try
to replicate some of them! And you can always ask me - I can certainly
help you navigate GHC a bit to get somewhere.

But there are still other things. The Win32 package for example, is
dreadfully lacking in maintainership. While we merge patches, it would
be great to see a Windows developer spearhead and clean it up - we
could even make some improvements in GHC itself based on this. This
would be an excellent opportunity to make a good impact in the broader
ecosystem!

Finally, we desperately need someone to consult with when we're up a
creek. Are certain patches OK for Windows? What's the best way to fix
certain bugs, or implement certain features? I feel like often we try
to think about this, but it's a bit lonely when nobody else is there
to help! I'm not sure how to fix this, other than encouraging things
like doing active code reviews and helping grind out some patches. But
at the very minimum, I'd just like to talk with you about things
perhaps!

So in summary - the work so far is grand, and we want to help you do
more! And I'm sure everyone can help - there's always so much to do
and so little time, we need to encourage it all we can.

As Simon says: Upward and Onward!

[1] http://haskell.inf.elte.hu/builders/
[2] https://ghc.haskell.org/trac/ghc/query?status=!closed&os=Windows&desc=1&order=id

--
Regards,

Austin Seipp, Haskell Consultant
Well-Typed LLP, http://www.well-typed.com/



--
Gintautas Miliauskas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/ghc-devs/attachments/20141028/e4fd1c96/attachment-0001.html>


More information about the ghc-devs mailing list