"stable" vs "security stable"

"stable" vs "security stable"

[Coywolf Qi Hunt]

Hello,

I find the kernel.org first page inconvenient for some people somehow
since the security stable came.

Now on the kernel.org page, we have 2.6.13.3 and 2.6.14-rc3. If one
wants to get 2.6.14-rc3, he shouldn't get 2.6.14-rc3 Full, but
2.6.14-rc3 Patch and 2.6.13 Full, which isn't there unfortunately. I
suggest we name 2.6.13.3 "security stable", and 2.6.13 "stable".

Fix the 1st line as:
s/The latest stable version/The latest security stable version/

Add a line below:
+The latest stable version of the Linux kernel is:  	2.6.13

And adjust the F,V on the right accordingly.

The rational behind is:
2.6.13.3 is security stable, which we suggest users to use. However
the stable is still 2.6.13.

Comments?
--
Coywolf Qi Hunt
http://sosdg.org/~coywolf/

Re: "stable" vs "security stable"

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 2624 bytes --]

On Sun, 09 Oct 2005 14:07:19 +0800, Coywolf Qi Hunt said:
> Hello,
> 
> I find the kernel.org first page inconvenient for some people somehow
> since the security stable came.

Unfortunately, it's a "stable", not "security stable" release.  Although
a large proportion of the fixes are security-related, the aren't *all*
security - there's also the occasional brown-bag bug or unexpected side
effect that simply causes incorrect operation of the kernel.

Having said that, Coywolf *is* right in that it's a bit unclear that
you have to fetch the 'F'(ull) 2.6.13.3, then get the patch, put that
on with patch -R to get a 2.6.13 tree, and then apply the 2.6.14-rc3 patch.
(Although if you realize that 14-rc3 is diffed off 13.0, not 13.3, it's not
that bad at all)...

I admit being torn between encouraging more people to try -rc kernels, and
wanting to enforce a minimum clue level on those trying to do so....

Hmm.. what if we did something like this:

diff -rup linux-2.6.13/dot.release linux-2.6.13.3/dot.release
--- linux-2.6.13/dot.release    2005-10-09 03:09:54.000000000 -0400
+++ linux-2.6.13.3/dot.release  2005-10-09 03:12:02.000000000 -0400
@@ -1,2 +1,2 @@
-This is a base release 2.6.13.  Stable patches, 2.6.14-rc patches,
-and the final 2.6.14 patch should be applied to this.
+This is a dot release. You need to patch -R the .3 patch before
+attempting to apply a .14-rc or .14 patch.

And then build the 14-rc3 patch:

diff -rup linux-2.6.13/dot.release linux-2.6.14-rc3/dot.release
--- linux-2.6.13/dot.release    2005-10-09 03:09:54.000000000 -0400
+++ linux-2.6.14-rc3/dot.release        2005-10-09 03:03:40.000000000 -0400
@@ -1,2 +1,3 @@
-This is a base release 2.6.13.  Stable patches, 2.6.14-rc patches,
-and the final 2.6.14 patch should be applied to this.
+This is a 14-rc3 release.  The patch will bomb out if you try
+to apply it to anything other than a 2.6.13.0 tree.  Did you
+remember to 'patch -R' any 2.6.13.N 'stable' patch first?

Now if we arrange for that to be the first diff in the patchfile, and
they get it wrong, they'll see:

% patch -p1 < 2.6.14-rc3.patch 
patching file dot.release
Hunk #1 FAILED at 1.
1 out of 1 hunk FAILED -- saving rejects to file dot.release.rej
% cat dot.release.rej 
***************
*** 1,2 ****
- This is a base release 2.6.13.  Stable patches, 2.6.14-rc patches,
- and the final 2.6.14 patch should be applied to this.
--- 1,3 ----
+ This is a 14-rc3 release.  The patch will bomb out if you try
+ to apply it to anything other than a 2.6.13.0 tree.  Did you
+ remember to 'patch -R' any 2.6.13.N 'stable' patch first?

(OK, it's a silly 3AM idea. ;)

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: "stable" vs "security stable"

[Coywolf Qi Hunt]

On 10/9/05, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:
> On Sun, 09 Oct 2005 14:07:19 +0800, Coywolf Qi Hunt said:
> > Hello,
> >
> > I find the kernel.org first page inconvenient for some people somehow
> > since the security stable came.
>
> Unfortunately, it's a "stable", not "security stable" release.  Although

It is "security stable". Let's take this new notation from now on.
"Security Stable" doesn't have to be all security related.

(you want stable@kernel.org to replace security@kernel.org too?)

> a large proportion of the fixes are security-related, the aren't *all*
> security - there's also the occasional brown-bag bug or unexpected side
> effect that simply causes incorrect operation of the kernel.
>
> Having said that, Coywolf *is* right in that it's a bit unclear that
> you have to fetch the 'F'(ull) 2.6.13.3, then get the patch, put that
> on with patch -R to get a 2.6.13 tree, and then apply the 2.6.14-rc3 patch.
> (Although if you realize that 14-rc3 is diffed off 13.0, not 13.3, it's not
> that bad at all)...
>
> I admit being torn between encouraging more people to try -rc kernels, and
> wanting to enforce a minimum clue level on those trying to do so....
>
> Hmm.. what if we did something like this:

What you did is so stupid to me to to use -R every time. -R implies
something wrong, and need to revert.

>
> diff -rup linux-2.6.13/dot.release linux-2.6.13.3/dot.release
> --- linux-2.6.13/dot.release    2005-10-09 03:09:54.000000000 -0400
> +++ linux-2.6.13.3/dot.release  2005-10-09 03:12:02.000000000 -0400
> @@ -1,2 +1,2 @@
> -This is a base release 2.6.13.  Stable patches, 2.6.14-rc patches,
> -and the final 2.6.14 patch should be applied to this.
> +This is a dot release. You need to patch -R the .3 patch before
> +attempting to apply a .14-rc or .14 patch.
>
> And then build the 14-rc3 patch:
>
> diff -rup linux-2.6.13/dot.release linux-2.6.14-rc3/dot.release
> --- linux-2.6.13/dot.release    2005-10-09 03:09:54.000000000 -0400
> +++ linux-2.6.14-rc3/dot.release        2005-10-09 03:03:40.000000000 -0400
> @@ -1,2 +1,3 @@
> -This is a base release 2.6.13.  Stable patches, 2.6.14-rc patches,
> -and the final 2.6.14 patch should be applied to this.
> +This is a 14-rc3 release.  The patch will bomb out if you try
> +to apply it to anything other than a 2.6.13.0 tree.  Did you
> +remember to 'patch -R' any 2.6.13.N 'stable' patch first?
>
> Now if we arrange for that to be the first diff in the patchfile, and
> they get it wrong, they'll see:
>
> % patch -p1 < 2.6.14-rc3.patch
> patching file dot.release
> Hunk #1 FAILED at 1.
> 1 out of 1 hunk FAILED -- saving rejects to file dot.release.rej
> % cat dot.release.rej
> ***************
> *** 1,2 ****
> - This is a base release 2.6.13.  Stable patches, 2.6.14-rc patches,
> - and the final 2.6.14 patch should be applied to this.
> --- 1,3 ----
> + This is a 14-rc3 release.  The patch will bomb out if you try
> + to apply it to anything other than a 2.6.13.0 tree.  Did you
> + remember to 'patch -R' any 2.6.13.N 'stable' patch first?
>
> (OK, it's a silly 3AM idea. ;)

(3pm here. :)
--
Coywolf Qi Hunt
http://sosdg.org/~coywolf/

Re: "stable" vs "security stable"

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 634 bytes --]

On Sun, 09 Oct 2005 15:44:38 +0800, Coywolf Qi Hunt said:
> On 10/9/05, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:

> It is "security stable". Let's take this new notation from now on.
> "Security Stable" doesn't have to be all security related.

Tell you what - you convince the -stable team, and I'll go along with it..

> (you want stable@kernel.org to replace security@kernel.org too?)

You're the one who called it "security stable" ;)

> What you did is so stupid to me to to use -R every time. -R implies
> something wrong, and need to revert.

Umm... my diff had *lower case* -r (recursive), not -R (revert)...


[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: "stable" vs "security stable"

[Coywolf Qi Hunt]

On 10/9/05, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:
> On Sun, 09 Oct 2005 15:44:38 +0800, Coywolf Qi Hunt said:
> > On 10/9/05, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:
>
> > It is "security stable". Let's take this new notation from now on.
> > "Security Stable" doesn't have to be all security related.
>
> Tell you what - you convince the -stable team, and I'll go along with it..
>

Better be "stable" and "base". 2.6.13.3 is the latest stable, 2.6.13
is the latest base.

> > (you want stable@kernel.org to replace security@kernel.org too?)
>
> You're the one who called it "security stable" ;)

My fault. I didn't realise stable@kernel.org exist, and CCed the wrong
list security@kernel.org.

>
> > What you did is so stupid to me to to use -R every time. -R implies
> > something wrong, and need to revert.
>
> Umm... my diff had *lower case* -r (recursive), not -R (revert)...

I mean `patch -R'.
--
Coywolf Qi Hunt
http://sosdg.org/~coywolf/

Re: "stable" vs "security stable"

[Bill Davidsen]

Coywolf Qi Hunt wrote:
> On 10/9/05, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:
> 
>>On Sun, 09 Oct 2005 15:44:38 +0800, Coywolf Qi Hunt said:
>>
>>>On 10/9/05, Valdis.Kletnieks@vt.edu <Valdis.Kletnieks@vt.edu> wrote:
>>
>>>It is "security stable". Let's take this new notation from now on.
>>>"Security Stable" doesn't have to be all security related.
>>
>>Tell you what - you convince the -stable team, and I'll go along with it..
>>
> 
> 
> Better be "stable" and "base". 2.6.13.3 is the latest stable, 2.6.13
> is the latest base.

I think the idea of having the most recent "base release," and -stable, 
and -rc, and -git, are desirable, with some clear terms. Those of us who 
started with ftp and never felt the need for using a GUI have long since 
learned what to keep and where to find it, but I bet most people use the 
web by now.

-- 
    -bill davidsen (davidsen@tmr.com)
"The secret to procrastination is to put things off until the
  last possible moment - but no longer"  -me

Re: [Security] "stable" vs "security stable"

[Chris Wright]

* Coywolf Qi Hunt (coywolf@gmail.com) wrote:
> I find the kernel.org first page inconvenient for some people somehow
> since the security stable came.

It's stable, not security stable.  It does contain security fixes
sometimes, but it is generally about patches that improve kernel
stability.

> Now on the kernel.org page, we have 2.6.13.3 and 2.6.14-rc3. If one
> wants to get 2.6.14-rc3, he shouldn't get 2.6.14-rc3 Full, but
> 2.6.14-rc3 Patch and 2.6.13 Full, which isn't there unfortunately. I
> suggest we name 2.6.13.3 "security stable", and 2.6.13 "stable".

Perhaps a column B for base.  Or just link to ketchup and be done with it.

Re: [Security] "stable" vs "security stable"

[Coywolf Qi Hunt]

On 10/9/05, Chris Wright <chrisw@osdl.org> wrote:
> * Coywolf Qi Hunt (coywolf@gmail.com) wrote:
> > I find the kernel.org first page inconvenient for some people somehow
> > since the security stable came.
>
> It's stable, not security stable.  It does contain security fixes
> sometimes, but it is generally about patches that improve kernel
> stability.

OK, "stable" vs "base" now. 2.6.13.3 is the latest stable, and 2.6.13
is the latest base.

>
> > Now on the kernel.org page, we have 2.6.13.3 and 2.6.14-rc3. If one
> > wants to get 2.6.14-rc3, he shouldn't get 2.6.14-rc3 Full, but
> > 2.6.14-rc3 Patch and 2.6.13 Full, which isn't there unfortunately. I
> > suggest we name 2.6.13.3 "security stable", and 2.6.13 "stable".
>
> Perhaps a column B for base.  Or just link to ketchup and be done with it.
>

--
Coywolf Qi Hunt
http://sosdg.org/~coywolf/