The new DTS language enhancements

The new DTS language enhancements

[Jon Loeliger]

Folks,

We have had some internal discussion inside Freescale
about my proposed DTS language enhancements now.
We feel that we have significant need for this approach,
and in the absence of a solid alternate proposal,
we would like to move forward with it.

Thanks,
jdl

Re: The new DTS language enhancements

[Kumar Gala]

On Oct 1, 2008, at 2:37 PM, Jon Loeliger wrote:

> Folks,
>
> We have had some internal discussion inside Freescale
> about my proposed DTS language enhancements now.
> We feel that we have significant need for this approach,
> and in the absence of a solid alternate proposal,
> we would like to move forward with it.

I agree with Jon here... if people aren't going to have useful  
comments on what changes need to be made to Jon's work we'd  
(Freescale) like to see a dtc-2.0 with these changes and getting those  
changes into the kernel for 2.6.28.

- k

Re: The new DTS language enhancements

[Grant Likely]

On Wed, Oct 1, 2008 at 3:11 PM, Kumar Gala <galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
>
> On Oct 1, 2008, at 2:37 PM, Jon Loeliger wrote:
>
>> Folks,
>>
>> We have had some internal discussion inside Freescale
>> about my proposed DTS language enhancements now.
>> We feel that we have significant need for this approach,
>> and in the absence of a solid alternate proposal,
>> we would like to move forward with it.
>
> I agree with Jon here... if people aren't going to have useful comments on
> what changes need to be made to Jon's work we'd (Freescale) like to see a
> dtc-2.0 with these changes and getting those changes into the kernel for
> 2.6.28.

Jon's changes are non-trivial and have long lasting implications.
First posting of the changes was made on Sept 23, just over a week
ago.  Jon told me about the new design only a month ago.  Not a lot of
time.

This is not the same as adding a new driver or supporting a new SoC.
The consequences have considerable broader reach.  Time must be
allowed for us to mull over the design.

Merging for 2.6.28 is too aggressive.

I appreciate that you're frustrated with the lack of alternate
proposals, but there hasn't been enough time yet.

g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.

Re: The new DTS language enhancements

[David Gibson]

On Wed, Oct 01, 2008 at 04:51:09PM -0600, Grant Likely wrote:
> On Wed, Oct 1, 2008 at 3:11 PM, Kumar Gala <galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
> >
> > On Oct 1, 2008, at 2:37 PM, Jon Loeliger wrote:
> >
> >> Folks,
> >>
> >> We have had some internal discussion inside Freescale
> >> about my proposed DTS language enhancements now.
> >> We feel that we have significant need for this approach,
> >> and in the absence of a solid alternate proposal,
> >> we would like to move forward with it.
> >
> > I agree with Jon here... if people aren't going to have useful comments on
> > what changes need to be made to Jon's work we'd (Freescale) like to see a
> > dtc-2.0 with these changes and getting those changes into the kernel for
> > 2.6.28.
> 
> Jon's changes are non-trivial and have long lasting implications.
> First posting of the changes was made on Sept 23, just over a week
> ago.  Jon told me about the new design only a month ago.  Not a lot of
> time.

I entirely agree, especially since the first I saw of the proposed
syntax was the implementation.

> This is not the same as adding a new driver or supporting a new SoC.
> The consequences have considerable broader reach.  Time must be
> allowed for us to mull over the design.

Indeed.  This will be syntax we have to live with for a long time, and
the current version does not sit at all well with me.

> Merging for 2.6.28 is too aggressive.
> 
> I appreciate that you're frustrated with the lack of alternate
> proposals, but there hasn't been enough time yet.

Yes.  I'm working on my own preferred approach now, as fast as I can
around other things I need to do.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

Re: The new DTS language enhancements

[Kumar Gala]

On Oct 1, 2008, at 7:42 PM, David Gibson wrote:

>>
>> Merging for 2.6.28 is too aggressive.
>>
>> I appreciate that you're frustrated with the lack of alternate
>> proposals, but there hasn't been enough time yet.
>
> Yes.  I'm working on my own preferred approach now, as fast as I can
> around other things I need to do.

Why not give some feedback or description of what you are thinking  
about?

- k

Re: The new DTS language enhancements

[Kumar Gala]

On Oct 1, 2008, at 5:51 PM, Grant Likely wrote:

> On Wed, Oct 1, 2008 at 3:11 PM, Kumar Gala  
> <galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
>>
>> On Oct 1, 2008, at 2:37 PM, Jon Loeliger wrote:
>>
>>> Folks,
>>>
>>> We have had some internal discussion inside Freescale
>>> about my proposed DTS language enhancements now.
>>> We feel that we have significant need for this approach,
>>> and in the absence of a solid alternate proposal,
>>> we would like to move forward with it.
>>
>> I agree with Jon here... if people aren't going to have useful  
>> comments on
>> what changes need to be made to Jon's work we'd (Freescale) like to  
>> see a
>> dtc-2.0 with these changes and getting those changes into the  
>> kernel for
>> 2.6.28.
>
> Jon's changes are non-trivial and have long lasting implications.
> First posting of the changes was made on Sept 23, just over a week
> ago.  Jon told me about the new design only a month ago.  Not a lot of
> time.
>
> This is not the same as adding a new driver or supporting a new SoC.
> The consequences have considerable broader reach.  Time must be
> allowed for us to mull over the design.
>
> Merging for 2.6.28 is too aggressive.
>
> I appreciate that you're frustrated with the lack of alternate
> proposals, but there hasn't been enough time yet.

I understand the short window, but if there isn't useful feedback what  
is one to do?  Just saying I don't like your solution and leaving at  
this isn't productive.  I'm all for getting this done right, but we  
need constructive feedback.

There are clearly some high level differences of opinions.  The only  
way I see of addressing those is by who is putting up code to address  
the problem versus not.. Jon's proposed a solution.. if no one else is  
(I know David's mentioned he's working on something) than it seems  
that wins by defacto.

I think David should be more transparent on what's he doing/thinking  
so we all can discuss it rather than us not knowing if/when he might  
produce something (and I've respond in such to him).

- k

Re: The new DTS language enhancements

[Josh Boyer]

On Wed, 1 Oct 2008 22:45:16 -0500
Kumar Gala <galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
> > This is not the same as adding a new driver or supporting a new SoC.
> > The consequences have considerable broader reach.  Time must be
> > allowed for us to mull over the design.
> >
> > Merging for 2.6.28 is too aggressive.
> >
> > I appreciate that you're frustrated with the lack of alternate
> > proposals, but there hasn't been enough time yet.
> 
> I understand the short window, but if there isn't useful feedback what  
> is one to do?  Just saying I don't like your solution and leaving at  
> this isn't productive.  I'm all for getting this done right, but we  
> need constructive feedback.
> 
> There are clearly some high level differences of opinions.  The only  
> way I see of addressing those is by who is putting up code to address  

And therein lies the rub.

A simple posting of grammar and syntax proposals before the code was
written would have probably led to better discussion and less time
wasted.  I'm all for seeing alternate proposals.  I just don't think
they need code implementation backing them to start with.

josh

Re: The new DTS language enhancements

[Kumar Gala]

On Oct 2, 2008, at 6:50 AM, Josh Boyer wrote:

> On Wed, 1 Oct 2008 22:45:16 -0500
> Kumar Gala <galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> wrote:
>>> This is not the same as adding a new driver or supporting a new SoC.
>>> The consequences have considerable broader reach.  Time must be
>>> allowed for us to mull over the design.
>>>
>>> Merging for 2.6.28 is too aggressive.
>>>
>>> I appreciate that you're frustrated with the lack of alternate
>>> proposals, but there hasn't been enough time yet.
>>
>> I understand the short window, but if there isn't useful feedback  
>> what
>> is one to do?  Just saying I don't like your solution and leaving at
>> this isn't productive.  I'm all for getting this done right, but we
>> need constructive feedback.
>>
>> There are clearly some high level differences of opinions.  The only
>> way I see of addressing those is by who is putting up code to address
>
> And therein lies the rub.
>
> A simple posting of grammar and syntax proposals before the code was
> written would have probably led to better discussion and less time
> wasted.  I'm all for seeing alternate proposals.  I just don't think
> they need code implementation backing them to start with.

I agree completely.. because if someone at least has a proposal than  
we have something that can be worked on.

- k

Re: The new DTS language enhancements

[David Gibson]

On Wed, Oct 01, 2008 at 10:45:16PM -0500, Kumar Gala wrote:
> On Oct 1, 2008, at 5:51 PM, Grant Likely wrote:
>> On Wed, Oct 1, 2008 at 3:11 PM, Kumar Gala <galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org> 
>> wrote:
[snip]
>> I appreciate that you're frustrated with the lack of alternate
>> proposals, but there hasn't been enough time yet.
>
> I understand the short window, but if there isn't useful feedback what  
> is one to do?  Just saying I don't like your solution and leaving at  
> this isn't productive.  I'm all for getting this done right, but we need 
> constructive feedback.
>
> There are clearly some high level differences of opinions.  The only way 
> I see of addressing those is by who is putting up code to address the 
> problem versus not.. Jon's proposed a solution.. if no one else is (I 
> know David's mentioned he's working on something) than it seems that wins 
> by defacto.
>
> I think David should be more transparent on what's he doing/thinking so 
> we all can discuss it rather than us not knowing if/when he might  
> produce something (and I've respond in such to him).

You're absolutely right.  I've been meaning to make some proposals
about where I think this should go, but it's been on the back-burner
while I've been working on other things.  Although, the one bit I did
post - the integer expressions support - was met with deafening
silence, which is a bit frustrating.

It's late here in .au and my brain's turning to mush, but first thing
tomorrow, I'll send out a brain dump of my thoughts on expression /
macro / whatever support.  Promise! It's not a complete proposed
syntax - which is one of the reasons I haven't posted already - just
thoughts, fragments, rationales and such.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

Re: The new DTS language enhancements

[Jon Loeliger]

On Fri, 2008-10-03 at 00:36 +1000, David Gibson wrote:
>  Although, the one bit I did
> post - the integer expressions support - was met with deafening
> silence, which is a bit frustrating.

On the contrary, by the time you had posted your patch,
I had a complete run-time evaluation environment running
in my development tree and it seemed a dramatic step backwards
to apply your patch and then work around it.

In fact, I did take the portion of your patch that was
very relevant and applied it.  It was the complete basis
for the beginnings of testing expression functionality.

Very useful, as it turned out.

jdl

Re: The new DTS language enhancements

[David Gibson]

On Thu, Oct 02, 2008 at 11:23:11AM -0500, Jon Loeliger wrote:
> On Fri, 2008-10-03 at 00:36 +1000, David Gibson wrote:
> >  Although, the one bit I did
> > post - the integer expressions support - was met with deafening
> > silence, which is a bit frustrating.
> 
> On the contrary, by the time you had posted your patch,
> I had a complete run-time evaluation environment running
> in my development tree and it seemed a dramatic step backwards
> to apply your patch and then work around it.

Heh.  Well, I'm glad I'm not the only one guilty of being
insufficiently transparent about what I'm working on.  Bear in mind
that when I posted the expression support patch, the last thing I'd
heard on the topic was "yeah, we've been talking about this for ages
and should probably do it...".

I'm not dead-set against a runtime evaluation environment, by the way,
although I'm certainly not fully convinced we need it.  But I do
loathe the current proposed syntax.  I guess one reason I like the
macro approach is it's less syntax that needs to be invented, and
therefore less opportunity for language design mistakes.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson