Re: CONFIG_NLS=y in 2.6.39 packages (Was Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy)

2011-05-10 Thread Ian Campbell
On Tue, 2011-05-10 at 03:58 +0100, Ben Hutchings wrote:
 On Fri, 2011-05-06 at 13:36 +0100, Ian Campbell wrote:
  On Fri, 2011-04-29 at 09:53 +0100, Ian Campbell wrote:
   On Fri, 2011-04-29 at 05:03 +0100, Ben Hutchings wrote:
On Thu, 2011-04-28 at 08:38 +0100, Ian Campbell wrote:
 On Sun, 2011-04-24 at 15:59 +0100, Ben Hutchings wrote:
  The '686-bigmem' flavour will be renamed to '686-pae'; support both 
  names.
  The '686' flavour will be removed, but we can still offer it if 
  available.
 
 Do you also have patches for d-i's kernel-wedge and the main installer
 build?

No, I'm not familiar with those.  I did the kernel installation
selection myself because I've worked on it before.

 Also debian-cd will need changes too. Since the latter two are
 mainly due to Xen support I can take care of that if you like, 
 although
 I think it will depend on the kernel actually being uploaded since I
 can't see an easy way to support both names in the main installer 
 build.

So it depends on when the d-i team decides to switch to 2.6.39 (or
later).  But you can get your patches ready before then.
   
   Correct. I'll look into it.
  
  While I was hacking up an updated kernel-wedge/linux-kernel-di-i386-2.6
  to use 2.6.39 to test this I noticed that nls_base is now built in
  instead of modular, despite CONFIG_NLS=m under debian/config. This is
  because various =y options (e.g. CONFIG_PCI_LABEL, CONFIG_JOLIET) now
  appear to select NLS...
 
 JOLIET has selected NLS since the beginning of git history (covering 3
 Debian releases so far), though it shouldn't.

It's a module anyhow or rather it is a boolean hidden behind ISO9660_FS
which is tristate=m so I think its select turns into a =m not a y.

CONFIG_PCI_LABEL seems to be the actual culprit and it is a hidden
boolean so there's no real choice, other than disabling DMI and ACPI
which is obviously a non-starter.

 Given how many things require NLS, though, we may as well explicitly make it 
 built-in.

ACK, I see you have made this change, thanks!

Ian.
-- 
Ian Campbell

I am currently transitioning to a new OpenPGP key, please see:
http://www.hellion.org.uk/key-transition-2011-04-27-2F6BCD59-to-79074FA8.txt

BOFH excuse #159:

Stubborn processes


signature.asc
Description: This is a digitally signed message part


Re: CONFIG_NLS=y in 2.6.39 packages (Was Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy)

2011-05-09 Thread Ben Hutchings
On Fri, 2011-05-06 at 13:36 +0100, Ian Campbell wrote:
 On Fri, 2011-04-29 at 09:53 +0100, Ian Campbell wrote:
  On Fri, 2011-04-29 at 05:03 +0100, Ben Hutchings wrote:
   On Thu, 2011-04-28 at 08:38 +0100, Ian Campbell wrote:
On Sun, 2011-04-24 at 15:59 +0100, Ben Hutchings wrote:
 The '686-bigmem' flavour will be renamed to '686-pae'; support both 
 names.
 The '686' flavour will be removed, but we can still offer it if 
 available.

Do you also have patches for d-i's kernel-wedge and the main installer
build?
   
   No, I'm not familiar with those.  I did the kernel installation
   selection myself because I've worked on it before.
   
Also debian-cd will need changes too. Since the latter two are
mainly due to Xen support I can take care of that if you like, although
I think it will depend on the kernel actually being uploaded since I
can't see an easy way to support both names in the main installer build.
   
   So it depends on when the d-i team decides to switch to 2.6.39 (or
   later).  But you can get your patches ready before then.
  
  Correct. I'll look into it.
 
 While I was hacking up an updated kernel-wedge/linux-kernel-di-i386-2.6
 to use 2.6.39 to test this I noticed that nls_base is now built in
 instead of modular, despite CONFIG_NLS=m under debian/config. This is
 because various =y options (e.g. CONFIG_PCI_LABEL, CONFIG_JOLIET) now
 appear to select NLS...

JOLIET has selected NLS since the beginning of git history (covering 3
Debian releases so far), though it shouldn't.  Given how many things
require NLS, though, we may as well explicitly make it built-in.

 Probably not a big deal, nls_base.o is ~5k, but thought I'd mention it.
 
 It just means that the nls-core-modules udeb goes away.

OK.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part


CONFIG_NLS=y in 2.6.39 packages (Was Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy)

2011-05-06 Thread Ian Campbell
On Fri, 2011-04-29 at 09:53 +0100, Ian Campbell wrote:
 On Fri, 2011-04-29 at 05:03 +0100, Ben Hutchings wrote:
  On Thu, 2011-04-28 at 08:38 +0100, Ian Campbell wrote:
   On Sun, 2011-04-24 at 15:59 +0100, Ben Hutchings wrote:
The '686-bigmem' flavour will be renamed to '686-pae'; support both 
names.
The '686' flavour will be removed, but we can still offer it if 
available.
   
   Do you also have patches for d-i's kernel-wedge and the main installer
   build?
  
  No, I'm not familiar with those.  I did the kernel installation
  selection myself because I've worked on it before.
  
   Also debian-cd will need changes too. Since the latter two are
   mainly due to Xen support I can take care of that if you like, although
   I think it will depend on the kernel actually being uploaded since I
   can't see an easy way to support both names in the main installer build.
  
  So it depends on when the d-i team decides to switch to 2.6.39 (or
  later).  But you can get your patches ready before then.
 
 Correct. I'll look into it.

While I was hacking up an updated kernel-wedge/linux-kernel-di-i386-2.6
to use 2.6.39 to test this I noticed that nls_base is now built in
instead of modular, despite CONFIG_NLS=m under debian/config. This is
because various =y options (e.g. CONFIG_PCI_LABEL, CONFIG_JOLIET) now
appear to select NLS...

Probably not a big deal, nls_base.o is ~5k, but thought I'd mention it.

It just means that the nls-core-modules udeb goes away.

Ian.
-- 
Ian Campbell
Current Noise: Napalm Death - De-evolution Ad Nauseum

I couldn't possibly fail to disagree with you less.


-- 
To UNSUBSCRIBE, email to debian-boot-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Archive: 
http://lists.debian.org/1304685363.26692.155.ca...@zakaz.uk.xensource.com



Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy

2011-04-29 Thread Ian Campbell
On Fri, 2011-04-29 at 05:03 +0100, Ben Hutchings wrote:
 On Thu, 2011-04-28 at 08:38 +0100, Ian Campbell wrote:
  On Sun, 2011-04-24 at 15:59 +0100, Ben Hutchings wrote:
   The '686-bigmem' flavour will be renamed to '686-pae'; support both names.
   The '686' flavour will be removed, but we can still offer it if available.
  
  Do you also have patches for d-i's kernel-wedge and the main installer
  build?
 
 No, I'm not familiar with those.  I did the kernel installation
 selection myself because I've worked on it before.
 
  Also debian-cd will need changes too. Since the latter two are
  mainly due to Xen support I can take care of that if you like, although
  I think it will depend on the kernel actually being uploaded since I
  can't see an easy way to support both names in the main installer build.
 
 So it depends on when the d-i team decides to switch to 2.6.39 (or
 later).  But you can get your patches ready before then.

Correct. I'll look into it.

Ian.

-- 
Ian Campbell

I am currently transitioning to a new OpenPGP key, please see:
http://www.hellion.org.uk/key-transition-2011-04-27-2F6BCD59-to-79074FA8.txt

Yow!  Are we laid back yet?


signature.asc
Description: This is a digitally signed message part


Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy

2011-04-28 Thread Ian Campbell
On Sun, 2011-04-24 at 15:59 +0100, Ben Hutchings wrote:
 The '686-bigmem' flavour will be renamed to '686-pae'; support both names.
 The '686' flavour will be removed, but we can still offer it if available.

Do you also have patches for d-i's kernel-wedge and the main installer
build? Also debian-cd will need changes too. Since the latter two are
mainly due to Xen support I can take care of that if you like, although
I think it will depend on the kernel actually being uploaded since I
can't see an easy way to support both names in the main installer build.

Ian.
-- 
Ian Campbell

I am currently transitioning to a new OpenPGP key, please see:
http://www.hellion.org.uk/key-transition-2011-04-27-2F6BCD59-to-79074FA8.txt

The avoidance of taxes is the only intellectual pursuit that carries any
reward.
-- John Maynard Keynes


signature.asc
Description: This is a digitally signed message part


Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy

2011-04-28 Thread Ben Hutchings
On Thu, 2011-04-28 at 08:38 +0100, Ian Campbell wrote:
 On Sun, 2011-04-24 at 15:59 +0100, Ben Hutchings wrote:
  The '686-bigmem' flavour will be renamed to '686-pae'; support both names.
  The '686' flavour will be removed, but we can still offer it if available.
 
 Do you also have patches for d-i's kernel-wedge and the main installer
 build?

No, I'm not familiar with those.  I did the kernel installation
selection myself because I've worked on it before.

 Also debian-cd will need changes too. Since the latter two are
 mainly due to Xen support I can take care of that if you like, although
 I think it will depend on the kernel actually being uploaded since I
 can't see an easy way to support both names in the main installer build.

So it depends on when the d-i team decides to switch to 2.6.39 (or
later).  But you can get your patches ready before then.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part


Re: [PATCH base-installer] Update i386 kernel selection for new flavours in wheezy

2011-04-24 Thread Christian PERRIER
Quoting Ben Hutchings (b...@decadent.org.uk):
 The '686-bigmem' flavour will be renamed to '686-pae'; support both names.
 The '686' flavour will be removed, but we can still offer it if available.

I uploaded b-i as part of a mass upload today, but will upload again
so that this important change is already supported in daily builds.

Thanks for your patch (which I applied blindly, I have to admit!).



signature.asc
Description: Digital signature