[gentoo-embedded] glibc-2 status

[gentoo-embedded] glibc-2 status

[Natanael Copa]

There were some kind of conflict some time ago related to NLS in
glibc-2. It resulted in glibc-2 got masked.

Anyone who know the status of glibc-2 in embedded?
Are the problems solved in uclibc-0.9.28?

Thanks!

-- 
Natanael Copa

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[solar]

On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> There were some kind of conflict some time ago related to NLS in
> glibc-2. It resulted in glibc-2 got masked.

You mean glib2 vs glibc? 


> Anyone who know the status of glibc-2 in embedded?

Assuming you mean glib2 it still remains at the same state. 
This is a sore spot for some of us being that the only problem 
moving forward is coming from another Gentoo developer.

> Are the problems solved in uclibc-0.9.28?

Yes & No.
.28 has better support and could probably handle it.
ie built in iconv() handling and basic locales handling.
It's just not very ideal or mature at all.


-- 
solar <solar@gentoo.org>
Gentoo Linux

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[Mike Frysinger]

On Tuesday 27 December 2005 13:02, solar wrote:
> On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > Anyone who know the status of glibc-2 in embedded?
>
> Assuming you mean glib2 it still remains at the same state.
> This is a sore spot for some of us being that the only problem
> moving forward is coming from another Gentoo developer.

considering said developer has been afk for quite a while, why dont i just go 
ahead and include the patch in our ebuilds (especially since upstream has 
accepted it)
-mike
-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[solar]

On Tue, 2005-12-27 at 15:03 -0500, Mike Frysinger wrote:
> On Tuesday 27 December 2005 13:02, solar wrote:
> > On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > > Anyone who know the status of glibc-2 in embedded?
> >
> > Assuming you mean glib2 it still remains at the same state.
> > This is a sore spot for some of us being that the only problem
> > moving forward is coming from another Gentoo developer.
> 
> considering said developer has been afk for quite a while, 

cough good.

> why dont i just go 
> ahead and include the patch in our ebuilds (especially since upstream has 
> accepted it)

I was unaware it was accepted for glib or even submitted for that
matter. But I like your idea.


-- 
solar <solar@gentoo.org>
Gentoo Linux

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[Peter S. Mazinger]

On Tue, 27 Dec 2005, solar wrote:

> On Tue, 2005-12-27 at 15:03 -0500, Mike Frysinger wrote:
> > On Tuesday 27 December 2005 13:02, solar wrote:
> > > On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > > > Anyone who know the status of glibc-2 in embedded?
> > >
> > > Assuming you mean glib2 it still remains at the same state.
> > > This is a sore spot for some of us being that the only problem
> > > moving forward is coming from another Gentoo developer.
> > 
> > considering said developer has been afk for quite a while, 
> 
> cough good.
> 
> > why dont i just go 
> > ahead and include the patch in our ebuilds (especially since upstream has 
> > accepted it)
> 
> I was unaware it was accepted for glib or even submitted for that
> matter. But I like your idea.

The patch was only accepted for gtk+, but for glib2 I am sure it won't 
ever be acceptable, because upstream the test for iconv() and nls support 
will stop the build, if they are both not found.

nls is no real problem, because it can be replaced by dummies, yvasilev 
made a better patch for iconv() replacement.

I see the future rather by enabling iconv() in uClibc (but not the 
nls/libintl/gettext stuff). That allows proper glib2 builds and everything 
that depends on it.
The nls part can be disabled/replaced by dummies in 
glib-2/glib/gi18n[-lib].h, that will allow to solve all glib2 dependant 
apps by:
sed -i 's:libintl.h:gi18n.h:' <file>

I have built like this all of emerge gnome

Peter

-- 
Peter S. Mazinger <ps dot m at gmx dot net>           ID: 0xA5F059F2
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08  BB6E C389 975E A5F0 59F2

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[Natanael Copa]

On man, 2006-01-02 at 16:21 +0100, Peter S. Mazinger wrote:
> On Tue, 27 Dec 2005, solar wrote:
> 
> > On Tue, 2005-12-27 at 15:03 -0500, Mike Frysinger wrote:
> > > On Tuesday 27 December 2005 13:02, solar wrote:
> > > > On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > > > > Anyone who know the status of glibc-2 in embedded?
> > > >
> > > > Assuming you mean glib2 it still remains at the same state.
> > > > This is a sore spot for some of us being that the only problem
> > > > moving forward is coming from another Gentoo developer.
> > > 
> > > considering said developer has been afk for quite a while, 
> > 
> > cough good.
> > 
> > > why dont i just go 
> > > ahead and include the patch in our ebuilds (especially since upstream has 
> > > accepted it)
> > 
> > I was unaware it was accepted for glib or even submitted for that
> > matter. But I like your idea.
> 
> The patch was only accepted for gtk+, but for glib2 I am sure it won't 
> ever be acceptable, because upstream the test for iconv() and nls support 
> will stop the build, if they are both not found.

Why cannot upstream glib2 support non iconv/nls?

> nls is no real problem, because it can be replaced by dummies, yvasilev 
> made a better patch for iconv() replacement.
> 
> I see the future rather by enabling iconv() in uClibc (but not the 
> nls/libintl/gettext stuff). That allows proper glib2 builds and everything 
> that depends on it.
> The nls part can be disabled/replaced by dummies in 
> glib-2/glib/gi18n[-lib].h, that will allow to solve all glib2 dependant 
> apps by:
> sed -i 's:libintl.h:gi18n.h:' <file>
> 
> I have built like this all of emerge gnome

wow... including firefox?


-- 
Natanael Copa

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[Peter S. Mazinger]

On Mon, 2 Jan 2006, Natanael Copa wrote:

> On man, 2006-01-02 at 16:21 +0100, Peter S. Mazinger wrote:
> > On Tue, 27 Dec 2005, solar wrote:
> > 
> > > On Tue, 2005-12-27 at 15:03 -0500, Mike Frysinger wrote:
> > > > On Tuesday 27 December 2005 13:02, solar wrote:
> > > > > On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > > > > > Anyone who know the status of glibc-2 in embedded?
> > > > >
> > > > > Assuming you mean glib2 it still remains at the same state.
> > > > > This is a sore spot for some of us being that the only problem
> > > > > moving forward is coming from another Gentoo developer.
> > > > 
> > > > considering said developer has been afk for quite a while, 
> > > 
> > > cough good.
> > > 
> > > > why dont i just go 
> > > > ahead and include the patch in our ebuilds (especially since upstream has 
> > > > accepted it)
> > > 
> > > I was unaware it was accepted for glib or even submitted for that
> > > matter. But I like your idea.
> > 
> > The patch was only accepted for gtk+, but for glib2 I am sure it won't 
> > ever be acceptable, because upstream the test for iconv() and nls support 
> > will stop the build, if they are both not found.
> 
> Why cannot upstream glib2 support non iconv/nls?

w/o iconv it is not possible anymore (they removed the "tolerant" way in 
glib2), nls would be no problem

> 
> > nls is no real problem, because it can be replaced by dummies, yvasilev 
> > made a better patch for iconv() replacement.
> > 
> > I see the future rather by enabling iconv() in uClibc (but not the 
> > nls/libintl/gettext stuff). That allows proper glib2 builds and everything 
> > that depends on it.
> > The nls part can be disabled/replaced by dummies in 
> > glib-2/glib/gi18n[-lib].h, that will allow to solve all glib2 dependant 
> > apps by:
> > sed -i 's:libintl.h:gi18n.h:' <file>
> > 
> > I have built like this all of emerge gnome
> 
> wow... including firefox?

both mozilla/firefox

Peter

-- 
Peter S. Mazinger <ps dot m at gmx dot net>           ID: 0xA5F059F2
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08  BB6E C389 975E A5F0 59F2

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[Natanael Copa]

On tir, 2005-12-27 at 15:03 -0500, Mike Frysinger wrote:
> On Tuesday 27 December 2005 13:02, solar wrote:
> > On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > > Anyone who know the status of glibc-2 in embedded?
> >
> > Assuming you mean glib2 it still remains at the same state.
> > This is a sore spot for some of us being that the only problem
> > moving forward is coming from another Gentoo developer.
> 
> considering said developer has been afk for quite a while, why dont i just go 
> ahead and include the patch in our ebuilds (especially since upstream has 
> accepted it)

I would appreciate if we could go forward, yes. But it would be good if
it could be done without making the "people issue" worse.

> -mike

-- 
Natanael Copa

-- 
gentoo-embedded@gentoo.org mailing list

Re: [gentoo-embedded] glibc-2 status

[Natanael Copa]

On tir, 2005-12-27 at 13:02 -0500, solar wrote:
> On Tue, 2005-12-27 at 14:36 +0100, Natanael Copa wrote:
> > There were some kind of conflict some time ago related to NLS in
> > glibc-2. It resulted in glibc-2 got masked.
> 
> You mean glib2 vs glibc?

ofcourse (sorry)


> 
> > Anyone who know the status of glibc-2 in embedded?
> 
> Assuming you mean glib2 it still remains at the same state. 
> This is a sore spot for some of us being that the only problem 
> moving forward is coming from another Gentoo developer.

I had the feeling that this is not a technical problem. (from the
bugzilla) It would be nice if it could be worked out.

> > Are the problems solved in uclibc-0.9.28?
> 
> Yes & No.
> .28 has better support and could probably handle it.
> ie built in iconv() handling and basic locales handling.
> It's just not very ideal or mature at all.


-- 
Natanael Copa

-- 
gentoo-embedded@gentoo.org mailing list