ketchup script and 3.0

ketchup script and 3.0

[Frank Kingswood]

Hi!

With Linux 3.0 approaching rapidly, is the ketchup script known to work?
It has this tempting <ver> parameter but I've always passed in "2.6" there.

Frank

Re: ketchup script and 3.0

[Matt Mackall]

On Tue, 2011-06-07 at 08:40 +0100, Frank Kingswood wrote:
> Hi!
> 
> With Linux 3.0 approaching rapidly, is the ketchup script known to work?
> It has this tempting <ver> parameter but I've always passed in "2.6" there.

It's on its third maintainer now and I've already forgotten the name of
the new guy.

-- 
Mathematics is the supreme nostalgia of our time.

Re: ketchup script and 3.0

[Stratos Psomadakis]

On 06/07/2011 03:53 PM, Matt Mackall wrote:
> On Tue, 2011-06-07 at 08:40 +0100, Frank Kingswood wrote:
>> Hi!
>>
>> With Linux 3.0 approaching rapidly, is the ketchup script known to work?
>> It has this tempting <ver> parameter but I've always passed in "2.6" there.
> It's on its third maintainer now and I've already forgotten the name of
> the new guy.
It's me :P

I uploaded the ketchup code at github, and added some code to handle 3.x
versions, but I've not tested it very much, so it's still in a separate
branch. [1]
You can check it out/test it, if you want. If it works without problems,
when linux-3.0 gets released, I'll tag a new version of ketchup and
notify distro maintainers to upgrade their packages.

Thanks! :)

[1] https://github.com/psomas/ketchup/tree/linux-3

-- 
Stratos Psomadakis
<psomas@ece.ntua.gr>

Re: ketchup script and 3.0

[Matt Mackall]

On Tue, 2011-06-07 at 22:48 +0300, Stratos Psomadakis wrote:
> On 06/07/2011 03:53 PM, Matt Mackall wrote:
> > On Tue, 2011-06-07 at 08:40 +0100, Frank Kingswood wrote:
> >> Hi!
> >>
> >> With Linux 3.0 approaching rapidly, is the ketchup script known to work?
> >> It has this tempting <ver> parameter but I've always passed in "2.6" there.
> > It's on its third maintainer now and I've already forgotten the name of
> > the new guy.
> It's me :P
> 
> I uploaded the ketchup code at github, and added some code to handle 3.x
> versions, but I've not tested it very much, so it's still in a separate
> branch. [1]
> You can check it out/test it, if you want. If it works without problems,
> when linux-3.0 gets released, I'll tag a new version of ketchup and
> notify distro maintainers to upgrade their packages.

I took a brief glance at your changes. You'll probably want to teach it
that 2.6.39++ == 3.0 so that people can seamlessly move back and forth
between the two ranges. This wasn't something that made sense across the
2.4/2.6 transition.

Oh, wait, maybe I've spotted the code for this.

Thinking ahead just a bit, it'd be nice if we could just declare in
advance that 3.9++ == 4.0. If we're going to bump the major number at
arbitrary points, that's the most obvious one. It's approximately 3
years out at the current rate which seems like a good pace. Then tools
like ketchup and other tools that handle these version numbers could
just do all this once.

Linus?

-- 
Mathematics is the supreme nostalgia of our time.

Re: ketchup script and 3.0

[Stratos Psomadakis]

On 06/07/2011 11:10 PM, Matt Mackall wrote:
> On Tue, 2011-06-07 at 22:48 +0300, Stratos Psomadakis wrote:
>> On 06/07/2011 03:53 PM, Matt Mackall wrote:
>>> On Tue, 2011-06-07 at 08:40 +0100, Frank Kingswood wrote:
>>>> Hi!
>>>>
>>>> With Linux 3.0 approaching rapidly, is the ketchup script known to work?
>>>> It has this tempting <ver> parameter but I've always passed in "2.6" there.
>>> It's on its third maintainer now and I've already forgotten the name of
>>> the new guy.
>> It's me :P
>>
>> I uploaded the ketchup code at github, and added some code to handle 3.x
>> versions, but I've not tested it very much, so it's still in a separate
>> branch. [1]
>> You can check it out/test it, if you want. If it works without problems,
>> when linux-3.0 gets released, I'll tag a new version of ketchup and
>> notify distro maintainers to upgrade their packages.
> I took a brief glance at your changes. You'll probably want to teach it
> that 2.6.39++ == 3.0 so that people can seamlessly move back and forth
> between the two ranges. This wasn't something that made sense across the
> 2.4/2.6 transition.
>
> Oh, wait, maybe I've spotted the code for this.
>
> Thinking ahead just a bit, it'd be nice if we could just declare in
> advance that 3.9++ == 4.0. If we're going to bump the major number at
> arbitrary points, that's the most obvious one. It's approximately 3
> years out at the current rate which seems like a good pace. Then tools
> like ketchup and other tools that handle these version numbers could
> just do all this once.
>
> Linus?
Yeap, that would be nice, indeed. Otherwise, ketchup code (and other
tools probably) will get uglier and uglier as major numbers advance, and
I made it look ugly already (although this is probably 'thinking way
into the future').

-- 
Stratos Psomadakis
<psomas@ece.ntua.gr>

Re: ketchup script and 3.0

[Matt Mackall]

On Tue, 2011-06-07 at 23:32 +0300, Stratos Psomadakis wrote:
> On 06/07/2011 11:10 PM, Matt Mackall wrote:
> > On Tue, 2011-06-07 at 22:48 +0300, Stratos Psomadakis wrote:
> >> On 06/07/2011 03:53 PM, Matt Mackall wrote:
> >>> On Tue, 2011-06-07 at 08:40 +0100, Frank Kingswood wrote:
> >>>> Hi!
> >>>>
> >>>> With Linux 3.0 approaching rapidly, is the ketchup script known to work?
> >>>> It has this tempting <ver> parameter but I've always passed in "2.6" there.
> >>> It's on its third maintainer now and I've already forgotten the name of
> >>> the new guy.
> >> It's me :P
> >>
> >> I uploaded the ketchup code at github, and added some code to handle 3.x
> >> versions, but I've not tested it very much, so it's still in a separate
> >> branch. [1]
> >> You can check it out/test it, if you want. If it works without problems,
> >> when linux-3.0 gets released, I'll tag a new version of ketchup and
> >> notify distro maintainers to upgrade their packages.
> > I took a brief glance at your changes. You'll probably want to teach it
> > that 2.6.39++ == 3.0 so that people can seamlessly move back and forth
> > between the two ranges. This wasn't something that made sense across the
> > 2.4/2.6 transition.
> >
> > Oh, wait, maybe I've spotted the code for this.
> >
> > Thinking ahead just a bit, it'd be nice if we could just declare in
> > advance that 3.9++ == 4.0. If we're going to bump the major number at
> > arbitrary points, that's the most obvious one. It's approximately 3
> > years out at the current rate which seems like a good pace. Then tools
> > like ketchup and other tools that handle these version numbers could
> > just do all this once.
> >
> > Linus?
> Yeap, that would be nice, indeed. Otherwise, ketchup code (and other
> tools probably) will get uglier and uglier as major numbers advance, and
> I made it look ugly already (although this is probably 'thinking way
> into the future').

Well it can in fact be managed with a list of transition points rather
than a cascade of if statements.

But my point is that if we have adhoc transitions, we will encounter the
"fix all the scripts and websites" pain at every transition. And tools
that are managed via distros and the like can literally take years to
get into the hands of users. It'd be nice if the copy of ketchup shipped
in <enterprise distro> just worked 3 years from now because 4.0 wasn't a
surprise.

-- 
Mathematics is the supreme nostalgia of our time.

Re: ketchup script and 3.0

[david]

On Tue, 7 Jun 2011, Matt Mackall wrote:

> But my point is that if we have adhoc transitions, we will encounter the
> "fix all the scripts and websites" pain at every transition. And tools
> that are managed via distros and the like can literally take years to
> get into the hands of users. It'd be nice if the copy of ketchup shipped
> in <enterprise distro> just worked 3 years from now because 4.0 wasn't a
> surprise.

if you special case 2.4->2.6, and make the default that 4.0 > 3.x, 5.0 > 
4.x, etc won't things 'just work' for the forseeable future?

David Lang

Re: ketchup script and 3.0

[Matt Mackall]

On Tue, 2011-06-07 at 13:50 -0700, david@lang.hm wrote:
> On Tue, 7 Jun 2011, Matt Mackall wrote:
> 
> > But my point is that if we have adhoc transitions, we will encounter the
> > "fix all the scripts and websites" pain at every transition. And tools
> > that are managed via distros and the like can literally take years to
> > get into the hands of users. It'd be nice if the copy of ketchup shipped
> > in <enterprise distro> just worked 3 years from now because 4.0 wasn't a
> > surprise.
> 
> if you special case 2.4->2.6, and make the default that 4.0 > 3.x, 5.0 > 
> 4.x, etc won't things 'just work' for the forseeable future?

No, because you sometimes want to know what 2.6.39++ is and what 3.0--
is.

For instance, to upgrade from 2.6.37.2 to 3.1.2, ketchup will want to
download, cache, and apply:

patch-2.6.37.2 (reversed!)
patch-2.6.38
patch-2.6.39
patch-3.0 ?? <- hopefully Linus will make a delta against 2.6.39!
patch-3.1
patch-3.1.2


-- 
Mathematics is the supreme nostalgia of our time.

Re: ketchup script and 3.0

[Stratos Psomadakis]

On 06/07/2011 11:59 PM, Matt Mackall wrote:
> On Tue, 2011-06-07 at 13:50 -0700, david@lang.hm wrote:
>> On Tue, 7 Jun 2011, Matt Mackall wrote:
>>
>>> But my point is that if we have adhoc transitions, we will encounter the
>>> "fix all the scripts and websites" pain at every transition. And tools
>>> that are managed via distros and the like can literally take years to
>>> get into the hands of users. It'd be nice if the copy of ketchup shipped
>>> in <enterprise distro> just worked 3 years from now because 4.0 wasn't a
>>> surprise.
>> if you special case 2.4->2.6, and make the default that 4.0 > 3.x, 5.0 > 
>> 4.x, etc won't things 'just work' for the forseeable future?
> No, because you sometimes want to know what 2.6.39++ is and what 3.0--
> is.
>
> For instance, to upgrade from 2.6.37.2 to 3.1.2, ketchup will want to
> download, cache, and apply:
>
> patch-2.6.37.2 (reversed!)
> patch-2.6.38
> patch-2.6.39
> patch-3.0 ?? <- hopefully Linus will make a delta against 2.6.39!
> patch-3.1
> patch-3.1.2
Well, if this is a ketchup-only issue (and there's no other need for
defining at which point major numbers are going to inc), we could find a
solution/workaround (to future-proof it, without the need of constant
updating of the code with every major release), but it'll probably be
uglier, and I think that what Matt proposed is better.

Anyway, when this issue is clarified, I'll try to fix the code, in order
to handle next major releases, without requiring code changes.

Thanks,

-- 
Stratos Psomadakis
<psomas@ece.ntua.gr>

Re: ketchup script and 3.0

[Randy Dunlap]

On Tue, June 7, 2011 2:22 pm, Stratos Psomadakis wrote:
> On 06/07/2011 11:59 PM, Matt Mackall wrote:
>> On Tue, 2011-06-07 at 13:50 -0700, david@lang.hm wrote:
>>> On Tue, 7 Jun 2011, Matt Mackall wrote:
>>>
>>>> But my point is that if we have adhoc transitions, we will encounter
>>>> the
>>>> "fix all the scripts and websites" pain at every transition. And tools
>>>> that are managed via distros and the like can literally take years to
>>>> get into the hands of users. It'd be nice if the copy of ketchup
>>>> shipped
>>>> in <enterprise distro> just worked 3 years from now because 4.0 wasn't
>>>> a
>>>> surprise.
>>> if you special case 2.4->2.6, and make the default that 4.0 > 3.x, 5.0
>>> >
>>> 4.x, etc won't things 'just work' for the forseeable future?
>> No, because you sometimes want to know what 2.6.39++ is and what 3.0--
>> is.
>>
>> For instance, to upgrade from 2.6.37.2 to 3.1.2, ketchup will want to
>> download, cache, and apply:
>>
>> patch-2.6.37.2 (reversed!)
>> patch-2.6.38
>> patch-2.6.39
>> patch-3.0 ?? <- hopefully Linus will make a delta against 2.6.39!
>> patch-3.1
>> patch-3.1.2
> Well, if this is a ketchup-only issue (and there's no other need for
> defining at which point major numbers are going to inc), we could find a
> solution/workaround (to future-proof it, without the need of constant
> updating of the code with every major release), but it'll probably be
> uglier, and I think that what Matt proposed is better.

It's also an issue for scripts/patch-kernel, unless I decide
to just kill (drop) it.

> Anyway, when this issue is clarified, I'll try to fix the code, in order
> to handle next major releases, without requiring code changes.



-- 
~Randy

Re: ketchup script and 3.0

[Florian Mickler]

On Wed, 08 Jun 2011 00:22:29 +0300
Stratos Psomadakis <psomas@ece.ntua.gr> wrote:

> On 06/07/2011 11:59 PM, Matt Mackall wrote:
> > On Tue, 2011-06-07 at 13:50 -0700, david@lang.hm wrote:
> >> On Tue, 7 Jun 2011, Matt Mackall wrote:
> >>
> >>> But my point is that if we have adhoc transitions, we will encounter the
> >>> "fix all the scripts and websites" pain at every transition. And tools
> >>> that are managed via distros and the like can literally take years to
> >>> get into the hands of users. It'd be nice if the copy of ketchup shipped
> >>> in <enterprise distro> just worked 3 years from now because 4.0 wasn't a
> >>> surprise.
> >> if you special case 2.4->2.6, and make the default that 4.0 > 3.x, 5.0 > 
> >> 4.x, etc won't things 'just work' for the forseeable future?
> > No, because you sometimes want to know what 2.6.39++ is and what 3.0--
> > is.
> >
> > For instance, to upgrade from 2.6.37.2 to 3.1.2, ketchup will want to
> > download, cache, and apply:
> >
> > patch-2.6.37.2 (reversed!)
> > patch-2.6.38
> > patch-2.6.39
> > patch-3.0 ?? <- hopefully Linus will make a delta against 2.6.39!
> > patch-3.1
> > patch-3.1.2
> Well, if this is a ketchup-only issue (and there's no other need for
> defining at which point major numbers are going to inc), we could find a
> solution/workaround (to future-proof it, without the need of constant
> updating of the code with every major release), but it'll probably be
> uglier, and I think that what Matt proposed is better.
> 
> Anyway, when this issue is clarified, I'll try to fix the code, in order
> to handle next major releases, without requiring code changes.
> 
> Thanks,
> 

Well, what I would do is to somehow get kernel.org to generate a
tag-list from git, append '-release' to the tag names which do not end
in '-rc[0-9]+' and then sort this list alphanumerical and just let
that be the first thing ketchup downloads. 

Regards,
Flo

Re: ketchup script and 3.0

[Américo Wang]

On Tue, Jun 07, 2011 at 02:33:53PM -0700, Randy Dunlap wrote:
>
>It's also an issue for scripts/patch-kernel, unless I decide
>to just kill (drop) it.
>

I think we should merge ketchup and then drop scripts/patch-kernel.

Re: ketchup script and 3.0

[Randy Dunlap]

On Sat, 11 Jun 2011 14:02:11 +0800 Américo Wang wrote:

> On Tue, Jun 07, 2011 at 02:33:53PM -0700, Randy Dunlap wrote:
> >
> >It's also an issue for scripts/patch-kernel, unless I decide
> >to just kill (drop) it.
> >
> 
> I think we should merge ketchup and then drop scripts/patch-kernel.

Sounds good to me.

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***