[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Areas of Emdebian to integrate into Debian



On Sun, 3 Apr 2011 20:56:07 +0100
Hector Oron <hector.oron@gmail.com> wrote:

> 2011/4/3 Neil Williams <codehelp@debian.org>:
> 
> > Have you had ANY indication that ftpmaster is even
> > slightly keen on such a division?
> 
> No, ftpmaster has not hinted me, but there is a real timing issue, and
> I am not sure it is ftpmasters call, but buildd and kernel team call,
> if it possible to do it and it improves Debian, why not?

I think you'll find it to be the ftpmaster who makes the call for what
actually gets into the ftp archive. buildd and kernel can put stuff
elsewhere but what gets onto ftp.debian.org is down to ftpmaster.

> Well, I was talking with Emdebian hat on. I do not think that is
> Emdebian business. 

Emdebian is frequently asked for kernels and we don't have the
resources to support more than one or two kernel packages for arm let
alone other architectures - cross-built or not.

The first thing new Emdebian users need when starting with Emdebian is
a kernel. It's always the first issue on IRC and on the list:

Q: How do I put Emdebian on this ARM board? 
A: Do you have a kernel?

You've seen that as often as I have.

We can make preparing the rootfs easier with multistrap additions but
at the end, it always comes down to getting a kernel.

Don't you think that if everyone combines patches we at least have a
chance of getting a starting point for people to use for their own
boards? It won't support every esoteric connection on the board
necessarily but at least it might get people started.

Tweaking an exising kernel is a lot more friendly than telling people
to Google for an entire kernel for every individual board and variant.

> With Balloon Board developer hat on, yes we could
> try to merge patches into mainline, and try to be careful when doing
> it with others platforms, which it is a time killer, that it is
> probably why those patches are not mainlined. Maybe Linaro can help
> out on that or that attempt Pengutronix did creating send-patches.org
> platform for such cases.

True.

Time is also lost when patches aren't prepared for mainline and have to
be manually updated for a newer kernel version which isn't as new as
mainline.

> > ... and encourage everyone talking about kernels or wanting kernels
> > from Emdebian to work as part of a larger community and collate patches.
> 
> Debian already does that, as being a requirement to merge patches into
> mainline tree. Again, what policy do we want to follow at Emdebian?

If we can, try to help get useable kernel packages into Debian for arm
to support Emdebian users? I don't know, I'm asking.

> > We don't have to do the work. We can encourage those who do the work to
> > try and make things go more smoothly.
> 
> On real world, time is money, and merging things properly takes time.
> Companies usually have deadlines, short time and lot of work to
> complete the goals. That is probably main blocker to submit/merge
> those patches into main tree.

Yes, time is short but the choice is: Should Emdebian people (not just
developers) spend time (often alone) porting old patches to new kernel
versions or spend (potentially less) time working with others to
combine patches into a form which can be suitable for mainline? It's
less time overall.

Do we have other objectives which don't involve kernels or is that the
main gap in my initial list?

-- 


Neil Williams
=============
http://www.linux.codehelp.co.uk/

Attachment: pgpBky2qEbWe7.pgp
Description: PGP signature


Reply to: