[Buildroot] svn commit: trunk/buildroot/target/device

[Buildroot] svn commit: trunk/buildroot/target/device

[aldot at uclibc.org]

Author: aldot
Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
New Revision: 19862

Log:
- Atmel mirrors are only of interrest for atmel


Modified:
   trunk/buildroot/target/device/Mirrors.in


Changeset:
Modified: trunk/buildroot/target/device/Mirrors.in
===================================================================
--- trunk/buildroot/target/device/Mirrors.in	2007-09-16 19:41:40 UTC (rev 19861)
+++ trunk/buildroot/target/device/Mirrors.in	2007-09-17 07:54:52 UTC (rev 19862)
@@ -4,14 +4,16 @@
 	string "Sourceforge mirror site"
 	default "easynews"
 	help
-	  Sourceforge has a system of mirror sites.  Some sites may be closer
-	  to your location, and sometimes mirror sites go down and are no longer
-	  available.  This option allows you to select your preferred Sourceforge
-	  mirror site.
+	  Sourceforge has a system of mirror sites.  Some sites may be
+	  closer to your location, and sometimes mirror sites go down
+	  and are no longer available.  This option allows you to select
+	  your preferred Sourceforge mirror site.
 
 	  The list of mirrors is available here:
 	  http://prdownloads.sourceforge.net/index-sf.html?download
 
+if BR2_avr32
 source	"target/device/Atmel/Mirrors.in"
+endif
 
 endmenu

[Buildroot] svn commit: trunk/buildroot/target/device

[Hans-Christian Egtvedt]

On Mon, 2007-09-17 at 00:55 -0700, aldot at uclibc.org wrote:
> Author: aldot
> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
> New Revision: 19862
> 
> Log:
> - Atmel mirrors are only of interrest for atmel
> 
> 
> Modified:
>    trunk/buildroot/target/device/Mirrors.in
> 
> 
> Changeset:
> Modified: trunk/buildroot/target/device/Mirrors.in
> ===================================================================
> --- trunk/buildroot/target/device/Mirrors.in	2007-09-16 19:41:40 UTC (rev 19861)
> +++ trunk/buildroot/target/device/Mirrors.in	2007-09-17 07:54:52 UTC (rev 19862)
> @@ -4,14 +4,16 @@
>  	string "Sourceforge mirror site"
>  	default "easynews"
>  	help
> -	  Sourceforge has a system of mirror sites.  Some sites may be closer
> -	  to your location, and sometimes mirror sites go down and are no longer
> -	  available.  This option allows you to select your preferred Sourceforge
> -	  mirror site.
> +	  Sourceforge has a system of mirror sites.  Some sites may be
> +	  closer to your location, and sometimes mirror sites go down
> +	  and are no longer available.  This option allows you to select
> +	  your preferred Sourceforge mirror site.
>  
>  	  The list of mirrors is available here:
>  	  http://prdownloads.sourceforge.net/index-sf.html?download
>  
> +if BR2_avr32

Add a "|| BR2_arm", as this is used for AT91 stuff AFAIK.

>  source	"target/device/Atmel/Mirrors.in"
> +endif
>  
>  endmenu

-- 
With kind regards,
Hans-Christian Egtvedt, Applications Engineer

[Buildroot] svn commit: trunk/buildroot/target/device

[Bernhard Fischer]

On Mon, Sep 17, 2007 at 10:03:11AM +0200, Hans-Christian Egtvedt wrote:
>On Mon, 2007-09-17 at 00:55 -0700, aldot at uclibc.org wrote:
>> Author: aldot
>> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
>> New Revision: 19862
>> 
>> Log:
>> - Atmel mirrors are only of interrest for atmel
>> 
>> 
>> Modified:
>>    trunk/buildroot/target/device/Mirrors.in
>> 
>> 
>> Changeset:
>> Modified: trunk/buildroot/target/device/Mirrors.in
>> ===================================================================
>> --- trunk/buildroot/target/device/Mirrors.in	2007-09-16 19:41:40 UTC (rev 19861)
>> +++ trunk/buildroot/target/device/Mirrors.in	2007-09-17 07:54:52 UTC (rev 19862)
>> @@ -4,14 +4,16 @@
>>  	string "Sourceforge mirror site"
>>  	default "easynews"
>>  	help
>> -	  Sourceforge has a system of mirror sites.  Some sites may be closer
>> -	  to your location, and sometimes mirror sites go down and are no longer
>> -	  available.  This option allows you to select your preferred Sourceforge
>> -	  mirror site.
>> +	  Sourceforge has a system of mirror sites.  Some sites may be
>> +	  closer to your location, and sometimes mirror sites go down
>> +	  and are no longer available.  This option allows you to select
>> +	  your preferred Sourceforge mirror site.
>>  
>>  	  The list of mirrors is available here:
>>  	  http://prdownloads.sourceforge.net/index-sf.html?download
>>  
>> +if BR2_avr32
>
>Add a "|| BR2_arm", as this is used for AT91 stuff AFAIK.

BR2_arm is inaccurate, BR2_TARGET_ATMEL or something like that, dunno.
>
>>  source	"target/device/Atmel/Mirrors.in"
>> +endif
>>  
>>  endmenu

[Buildroot] svn commit: trunk/buildroot/target/device

[Bernhard Fischer]

On Mon, Sep 17, 2007 at 12:43:22PM +0200, Ulf Samuelsson wrote:
>m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
>> Author: aldot
>> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
>> New Revision: 19862
>> 
>> Log:
>> - Atmel mirrors are only of interrest for atmel
>> 
>> 
>> Modified:
>>    trunk/buildroot/target/device/Mirrors.in
>> 
>> 
>
>And making it depending on BR2_ATMEL or similar makes it impossible to 
>put package support on that ftp site, so don't create any dependency
>at all.

This can depend on BR2_TARGET_ATMEL or some at91 config symbol.
Depending on BR2_arm is wrong since i want to build arm toolchains but
use upstream sources, not some vendor supplied tarballs.

If we need to mirror certain packages, we can of course do this (as the
buildroot project), but using random vendor's locked-down ftp server for
this purpose is not a sensible thing to do, IMO.

[Buildroot] svn commit: trunk/buildroot/target/device

[Ulf Samuelsson]

m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
> Author: aldot
> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
> New Revision: 19862
> 
> Log:
> - Atmel mirrors are only of interrest for atmel
> 
> 
You just broke AT91 ARM support.
Please revert!



> Modified:
>    trunk/buildroot/target/device/Mirrors.in
> 
> 
> Changeset:
> Modified: trunk/buildroot/target/device/Mirrors.in
> ===================================================================
> --- trunk/buildroot/target/device/Mirrors.in	2007-09-16 19:41:40 UTC (rev 19861)
> +++ trunk/buildroot/target/device/Mirrors.in	2007-09-17 07:54:52 UTC (rev 19862)
> @@ -4,14 +4,16 @@
>  	string "Sourceforge mirror site"
>  	default "easynews"
>  	help
> -	  Sourceforge has a system of mirror sites.  Some sites may be closer
> -	  to your location, and sometimes mirror sites go down and are no longer
> -	  available.  This option allows you to select your preferred Sourceforge
> -	  mirror site.
> +	  Sourceforge has a system of mirror sites.  Some sites may be
> +	  closer to your location, and sometimes mirror sites go down
> +	  and are no longer available.  This option allows you to select
> +	  your preferred Sourceforge mirror site.
>  
>  	  The list of mirrors is available here:
>  	  http://prdownloads.sourceforge.net/index-sf.html?download
>  
> +if BR2_avr32
>  source	"target/device/Atmel/Mirrors.in"
> +endif
>  
>  endmenu
> 
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot

[Buildroot] svn commit: trunk/buildroot/target/device

[Ulf Samuelsson]

m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
> Author: aldot
> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
> New Revision: 19862
> 
> Log:
> - Atmel mirrors are only of interrest for atmel
> 
> 
> Modified:
>    trunk/buildroot/target/device/Mirrors.in
> 
> 

And making it depending on BR2_ATMEL or similar makes it impossible to 
put package support on that ftp site, so don't create any dependency
at all.


> Changeset:
> Modified: trunk/buildroot/target/device/Mirrors.in
> ===================================================================
> --- trunk/buildroot/target/device/Mirrors.in	2007-09-16 19:41:40 UTC (rev 19861)
> +++ trunk/buildroot/target/device/Mirrors.in	2007-09-17 07:54:52 UTC (rev 19862)
> @@ -4,14 +4,16 @@
>  	string "Sourceforge mirror site"
>  	default "easynews"
>  	help
> -	  Sourceforge has a system of mirror sites.  Some sites may be closer
> -	  to your location, and sometimes mirror sites go down and are no longer
> -	  available.  This option allows you to select your preferred Sourceforge
> -	  mirror site.
> +	  Sourceforge has a system of mirror sites.  Some sites may be
> +	  closer to your location, and sometimes mirror sites go down
> +	  and are no longer available.  This option allows you to select
> +	  your preferred Sourceforge mirror site.
>  
>  	  The list of mirrors is available here:
>  	  http://prdownloads.sourceforge.net/index-sf.html?download
>  
> +if BR2_avr32
>  source	"target/device/Atmel/Mirrors.in"
> +endif
>  
>  endmenu
> 
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot

[Buildroot] svn commit: trunk/buildroot/target/device

[Bernhard Fischer]

On Mon, Sep 17, 2007 at 02:16:44PM +0200, Ulf Samuelsson wrote:
>m?n 2007-09-17 klockan 11:27 +0200 skrev Bernhard Fischer:
>> On Mon, Sep 17, 2007 at 12:43:22PM +0200, Ulf Samuelsson wrote:
>> >m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
>> >> Author: aldot
>> >> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
>> >> New Revision: 19862
>> >> 
>> >> Log:
>> >> - Atmel mirrors are only of interrest for atmel
>> >> 
>> >> 
>> >> Modified:
>> >>    trunk/buildroot/target/device/Mirrors.in
>> >> 
>> >> 
>> >
>> >And making it depending on BR2_ATMEL or similar makes it impossible to 
>> >put package support on that ftp site, so don't create any dependency
>> >at all.
>> 
>> This can depend on BR2_TARGET_ATMEL or some at91 config symbol.
>> Depending on BR2_arm is wrong since i want to build arm toolchains but
>> use upstream sources, not some vendor supplied tarballs.
>> 
>You asssume that there is never going to be something on the ftp site
>which is interesting to x86/ppc etc. users.
>That assumption is wrong, so therefore the dependency is wrong

You seem to be assuming that there is or will be something on that ftp
site that is interresting. What would that be? I don't miss alot of
stuff and if so, I somehow doubt that the missing bits would live on
an atmel tfp-server.

I ask you to fix the mirror-selection to reflect that a vendor may
provide additional patches on their site for their specific products for
a short period of time -- i.e. until the fixes are applied to upstream
sources -- but _not_ to pollute the rest.

Thank you.

[Buildroot] svn commit: trunk/buildroot/target/device

[Ulf Samuelsson]

m?n 2007-09-17 klockan 11:27 +0200 skrev Bernhard Fischer:
> On Mon, Sep 17, 2007 at 12:43:22PM +0200, Ulf Samuelsson wrote:
> >m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
> >> Author: aldot
> >> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
> >> New Revision: 19862
> >> 
> >> Log:
> >> - Atmel mirrors are only of interrest for atmel
> >> 
> >> 
> >> Modified:
> >>    trunk/buildroot/target/device/Mirrors.in
> >> 
> >> 
> >
> >And making it depending on BR2_ATMEL or similar makes it impossible to 
> >put package support on that ftp site, so don't create any dependency
> >at all.
> 
> This can depend on BR2_TARGET_ATMEL or some at91 config symbol.
> Depending on BR2_arm is wrong since i want to build arm toolchains but
> use upstream sources, not some vendor supplied tarballs.
> 
You asssume that there is never going to be something on the ftp site
which is interesting to x86/ppc etc. users.
That assumption is wrong, so therefore the dependency is wrong

/Ulf

> If we need to mirror certain packages, we can of course do this (as the
> buildroot project), but using random vendor's locked-down ftp server for
> this purpose is not a sensible thing to do, IMO.

[Buildroot] svn commit: trunk/buildroot/target/device

[Bernhard Fischer]

On Mon, Sep 17, 2007 at 04:56:18PM +0200, Ulf Samuelsson wrote:
>m?n 2007-09-17 klockan 13:10 +0200 skrev Bernhard Fischer:
>> On Mon, Sep 17, 2007 at 02:16:44PM +0200, Ulf Samuelsson wrote:
>> >m?n 2007-09-17 klockan 11:27 +0200 skrev Bernhard Fischer:
>> >> On Mon, Sep 17, 2007 at 12:43:22PM +0200, Ulf Samuelsson wrote:
>> >> >m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
>> >> >> Author: aldot
>> >> >> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
>> >> >> New Revision: 19862
>> >> >> 
>> >> >> Log:
>> >> >> - Atmel mirrors are only of interrest for atmel
>> >> >> 
>> >> >> 
>> >> >> Modified:
>> >> >>    trunk/buildroot/target/device/Mirrors.in
>> >> >> 
>> >> >> 
>> >> >
>> >> >And making it depending on BR2_ATMEL or similar makes it impossible to 
>> >> >put package support on that ftp site, so don't create any dependency
>> >> >at all.
>> >> 
>> >> This can depend on BR2_TARGET_ATMEL or some at91 config symbol.
>> >> Depending on BR2_arm is wrong since i want to build arm toolchains but
>> >> use upstream sources, not some vendor supplied tarballs.
>> >> 
>> >You asssume that there is never going to be something on the ftp site
>> >which is interesting to x86/ppc etc. users.
>> >That assumption is wrong, so therefore the dependency is wrong
>> 
>> You seem to be assuming that there is or will be something on that ftp
>> site that is interresting. What would that be?
>
>There are a couple of packages, that I plan to add at a later stage.

I will revert the wrong include until then, so you can fix that if it's
needed. Whether it is BR2_avr32||BR2_at91 or BR2_TARGET_ATMEL is
something that i leave up to you.
>
>>
>>  I don't miss alot of
>> stuff and if so, I somehow doubt that the missing bits would live on
>> an atmel tfp-server.
>> 
>> I ask you to fix the mirror-selection to reflect that a vendor may
>> provide additional patches on their site for their specific products for
>> a short period of time -- i.e. until the fixes are applied to upstream
>> sources -- but _not_ to pollute the rest.
>> 
>
>Exactly what problem are you trying to resolve?

There is no point in having any atmel mirror in my .config unless i
build for at91 or avr32 and choose to use an atmel toolchain or atmel
patches.

[Buildroot] svn commit: trunk/buildroot/target/device

[Ulf Samuelsson]

m?n 2007-09-17 klockan 13:10 +0200 skrev Bernhard Fischer:
> On Mon, Sep 17, 2007 at 02:16:44PM +0200, Ulf Samuelsson wrote:
> >m?n 2007-09-17 klockan 11:27 +0200 skrev Bernhard Fischer:
> >> On Mon, Sep 17, 2007 at 12:43:22PM +0200, Ulf Samuelsson wrote:
> >> >m?n 2007-09-17 klockan 00:55 -0700 skrev aldot at uclibc.org:
> >> >> Author: aldot
> >> >> Date: 2007-09-17 00:54:52 -0700 (Mon, 17 Sep 2007)
> >> >> New Revision: 19862
> >> >> 
> >> >> Log:
> >> >> - Atmel mirrors are only of interrest for atmel
> >> >> 
> >> >> 
> >> >> Modified:
> >> >>    trunk/buildroot/target/device/Mirrors.in
> >> >> 
> >> >> 
> >> >
> >> >And making it depending on BR2_ATMEL or similar makes it impossible to 
> >> >put package support on that ftp site, so don't create any dependency
> >> >at all.
> >> 
> >> This can depend on BR2_TARGET_ATMEL or some at91 config symbol.
> >> Depending on BR2_arm is wrong since i want to build arm toolchains but
> >> use upstream sources, not some vendor supplied tarballs.
> >> 
> >You asssume that there is never going to be something on the ftp site
> >which is interesting to x86/ppc etc. users.
> >That assumption is wrong, so therefore the dependency is wrong
> 
> You seem to be assuming that there is or will be something on that ftp
> site that is interresting. What would that be?

There are a couple of packages, that I plan to add at a later stage.

>
>  I don't miss alot of
> stuff and if so, I somehow doubt that the missing bits would live on
> an atmel tfp-server.
> 
> I ask you to fix the mirror-selection to reflect that a vendor may
> provide additional patches on their site for their specific products for
> a short period of time -- i.e. until the fixes are applied to upstream
> sources -- but _not_ to pollute the rest.
> 

Exactly what problem are you trying to resolve?


> Thank you.

[Buildroot] svn commit: trunk/buildroot/target/device

[Ulf Samuelsson]

> >Exactly what problem are you trying to resolve?
> 
> There is no point in having any atmel mirror in my .config unless i
> build for at91 or avr32 and choose to use an atmel toolchain or atmel
> patches.

if/when it will mirror all the tarballs used by Buildroot
it will be useful to everyone.
I am lobbying for this to happen.

/BR
Ulf Samuelsson

[Buildroot] local mirror management (bug #572)

[Bernhard Fischer]

On Tue, Sep 18, 2007 at 08:54:49AM +0200, Ulf Samuelsson wrote:
>
>> >Exactly what problem are you trying to resolve?
>> 
>> There is no point in having any atmel mirror in my .config unless i
>> build for at91 or avr32 and choose to use an atmel toolchain or atmel
>> patches.
>
>if/when it will mirror all the tarballs used by Buildroot
>it will be useful to everyone.
>I am lobbying for this to happen.

No, this is unacceptable. If you want to do stuff like that, look at the
"local mirror" patch in mantis (or the ML archives). See 1).

With the patch mentioned above, you (i.e. the user) is free to setup and
use any mirror she likes, be it the
http://buildroot.uclibc.org/downloads/buildroot-sources/
or another, potentially local directory (i.e. nothing you can talk to
via http or ftp).

We will not accept to rely on a random vendor's ftp collection, this is
conceptually broken since we have no influence on their ftp server and
the packages they may choose to have there or not.

1) http://bugs.busybox.net/view.php?id=572
has a rough draft that IIRC needs some cleanup and that should be
discussed in it's cleaned up version on this list.

[Buildroot] local mirror management (bug #572)

[Bernhard Fischer]

On Wed, Sep 19, 2007 at 11:53:39PM +0200, Ulf Samuelsson wrote:
>ons 2007-09-19 klockan 10:21 +0200 skrev Bernhard Fischer:

>> With the patch mentioned above, you (i.e. the user) is free to setup and
>> use any mirror she likes, be it the
>> http://buildroot.uclibc.org/downloads/buildroot-sources/

I've dropped a couple of "our" package-sources there, fyi.

>> or another, potentially local directory (i.e. nothing you can talk to
>> via http or ftp).

>> 1) http://bugs.busybox.net/view.php?id=572
>> has a rough draft that IIRC needs some cleanup and that should be
>> discussed in it's cleaned up version on this list.
>
>Don't be so negative.

I'm not negative (just look at the patch to see what i mean).

[Buildroot] local mirror management (bug #572)

[Ulf Samuelsson]

ons 2007-09-19 klockan 10:21 +0200 skrev Bernhard Fischer:
> On Tue, Sep 18, 2007 at 08:54:49AM +0200, Ulf Samuelsson wrote:
> >
> >> >Exactly what problem are you trying to resolve?
> >> 
> >> There is no point in having any atmel mirror in my .config unless i
> >> build for at91 or avr32 and choose to use an atmel toolchain or atmel
> >> patches.
> >
> >if/when it will mirror all the tarballs used by Buildroot
> >it will be useful to everyone.
> >I am lobbying for this to happen.
> 
> No, this is unacceptable. If you want to do stuff like that, look at the
> "local mirror" patch in mantis (or the ML archives). See 1).
> 
> With the patch mentioned above, you (i.e. the user) is free to setup and
> use any mirror she likes, be it the
> http://buildroot.uclibc.org/downloads/buildroot-sources/
> or another, potentially local directory (i.e. nothing you can talk to
> via http or ftp).
> 
> We will not accept to rely on a random vendor's ftp collection, this is
> conceptually broken since we have no influence on their ftp server and
> the packages they may choose to have there or not.
> 
> 1) http://bugs.busybox.net/view.php?id=572
> has a rough draft that IIRC needs some cleanup and that should be
> discussed in it's cleaned up version on this list.

Don't be so negative.
Any patch regarding mirrors will need to maintain a list of mirror
locations.

All packages should normally be downloaded from their current location
and the mirrors should only be used if the packet is not available
at the current location for some reason.

Best Regards
Ulf Samuelsson

[Buildroot] local mirror management (bug #572)

[Ulf Samuelsson]

ons 2007-09-19 klockan 23:20 +0200 skrev Bernhard Fischer:
> On Wed, Sep 19, 2007 at 11:53:39PM +0200, Ulf Samuelsson wrote:
> >ons 2007-09-19 klockan 10:21 +0200 skrev Bernhard Fischer:
> 
> >> With the patch mentioned above, you (i.e. the user) is free to setup and
> >> use any mirror she likes, be it the
> >> http://buildroot.uclibc.org/downloads/buildroot-sources/
> 
> I've dropped a couple of "our" package-sources there, fyi.
> 
> >> or another, potentially local directory (i.e. nothing you can talk to
> >> via http or ftp).
> 
> >> 1) http://bugs.busybox.net/view.php?id=572
> >> has a rough draft that IIRC needs some cleanup and that should be
> >> discussed in it's cleaned up version on this list.
> >
> >Don't be so negative.
> 
> I'm not negative (just look at the patch to see what i mean).




I don't see any conflict in these patches.
It is similar to what I had in mind.

You would define BR2_LOCAL_MIRROR_SITE as $(ATMEL_MIRROR)
It is not good to define it as the current value of ATMEL_MIRROR
since the site will eventually change.

Best Regards

[Buildroot] local mirror management (bug #572)

[Elizabeth Oldham]

Bernhard Fischer wrote:

> "local mirror" patch in mantis (or the ML archives). See 1).
> With the patch mentioned above, you (i.e. the user) is free to setup and
> use any mirror she likes, be it the
> http://buildroot.uclibc.org/downloads/buildroot-sources/
> or another, potentially local directory (i.e. nothing you can talk to
> via http or ftp).
> 
> 1) http://bugs.busybox.net/view.php?id=572

I'll bite, this is exactly what I need... I tweaked the patch in that 
bug and attach my version for comment. I dont know what clean up you had 
in mind but I'm happy to make further attempts.

Beth
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: local-mirror.patch
Url: http://busybox.net/lists/buildroot/attachments/20070920/fd87a513/attachment.diff 

[Buildroot] local mirror management (bug #572)

[Bernhard Fischer]

On Thu, Sep 20, 2007 at 10:38:52PM +0100, Elizabeth Oldham wrote:
> Bernhard Fischer wrote:
>
>> "local mirror" patch in mantis (or the ML archives). See 1).
>> With the patch mentioned above, you (i.e. the user) is free to setup and
>> use any mirror she likes, be it the
>> http://buildroot.uclibc.org/downloads/buildroot-sources/
>> or another, potentially local directory (i.e. nothing you can talk to
>> via http or ftp).
>> 1) http://bugs.busybox.net/view.php?id=572
>
> I'll bite, this is exactly what I need... I tweaked the patch in that bug 
> and attach my version for comment. I dont know what clean up you had in 
> mind but I'm happy to make further attempts.
>
> Beth

>Index: Config.in
>===================================================================
>--- Config.in	(revision 19922)
>+++ Config.in	(working copy)
>@@ -672,6 +672,39 @@
> 	string "Wget command"
> 	default "wget --passive-ftp -nd"
> 
>+config BR2_LOCAL_MIRROR
>+	bool "Use a local mirror"
>+      	default n

Please use just one leading tab. The help-text is usually like this:
<tab>help
<tab><space><space>description that fits in about 70 columns
Same for the majority below.
>+
>+config BR2_LOCAL_MIRROR_STRICT
>+	bool "Only download from mirror"
>+	default n
>+
>+config BR2_LOCAL_MIRROR_SITE
>+	string "Local mirror site:"
>+       	default "http://localhost/"
>+       	depends on BR2_LOCAL_MIRROR
>+
>+config BR2_LOCAL_MIRROR_LOCALHOSTS
>+      	string "Local hosts"
>+      	default ""
>+      	depends on BR2_LOCAL_MIRROR
>+      	help
>+          Do not change URL of specified hosts. Useful if you maintain
>+          your own packages.
>+
>+config BR2_LOCAL_MIRROR_GEN
>+      	bool "Update mirror build directory"
>+      	depends on !BR2_LOCAL_MIRROR_STRICT
>+      	help
>+          Update your mirror build directory with the packages that 
>+	  cannot be fetched from the mirror.
>+
>+config BR2_LOCAL_MIRROR_GENDIR
>+	string "mirror build directory"
>+      	depends on BR2_LOCAL_MIRROR_GEN
>+      	default "$(BASE_DIR)/mirror"
>+
> config BR2_SVN_CO
> 	string "Subversion (svn) command to download source tree"
> 	default "svn co"
>Index: package/Makefile.in
>===================================================================
>--- package/Makefile.in	(revision 19922)
>+++ package/Makefile.in	(working copy)
>@@ -62,6 +62,17 @@
> endif
> #########################################################################
> 
>+ifeq ($(BR2_LOCAL_MIRROR),y)
>+WGET:=MIRROR_WGET="$(strip $(subst ",, $(BR2_WGET)))" \
>+       MIRROR_SITE="$(strip $(subst ",, $(BR2_LOCAL_MIRROR_SITE)))" \
>+       MIRROR_LOCALHOSTS="$(strip $(subst ",, $(BR2_LOCAL_MIRROR_LOCALHOSTS)))" \
>+       MIRROR_STRICT="$(strip $(subst ",, $(BR2_LOCAL_MIRROR_STRICT)))" \
>+       MIRROR_GEN="$(strip $(subst ",, $(BR2_LOCAL_MIRROR_GEN)))" \
>+       MIRROR_GENDIR="$(strip $(subst ",, $(BR2_LOCAL_MIRROR_GENDIR)))" \
>+       $(BASE_DIR)/misc/mirror.sh

This will break vim's syntax highlighting, unfortunately. It needs
#"))
here. Fire up vim and :syntax on
>+else
>+WGET:=$(strip $(subst ",, $(BR2_WGET)))
#"))
>+endif
> 
> ifeq ($(BR2_TOOLCHAIN_BUILDROOT),y)
> TOOL_BUILD_DIR=$(BASE_DIR)/$(TOPDIR_PREFIX)toolchain_build_$(ARCH)$(ARCH_FPU_SUFFIX)$(TOPDIR_SUFFIX)
>Index: misc/mirror.sh

Please move that somewhere to toolchain/.

>===================================================================
>--- misc/mirror.sh	(revision 0)
>+++ misc/mirror.sh	(revision 0)
>@@ -0,0 +1,61 @@
>+#!/bin/bash

bash is not acceptable. It has to run on a POSIX shell if it can't be
done in make (which is clearly to be preferred).

>+
>+IFS=" "
>+
>+OPTIONS=""
>+USE_LOCAL=0
>+
>+for i in "$@"
>+do
>+  echo $i | grep '://' > /dev/null;
>+
>+  if [ "$?" = 0 ]
>+  then
>+    EXT_URL=$i
>+    HOST=`echo $i |sed 's+^.*://++' |sed 's+/.*++'`

You'll have to pass SED=$(XSED) down to that script (which shouldn't
exist in the first place, if possible) since
a) The host may not have sed installed
b) The host sed may be buggy
c) if one of a) b), we built a proper version for our own use.
>+    
>+    for j in $MIRROR_LOCAL_HOSTS
>+    do
>+      if [ "$j" = "$HOST" ]

Several test(1) impls will get this wrong if one of them is empty.
Prefix that with a char, like [ "x$j" = "x$HOST" ]

>+      then
>+        USE_LOCAL=1;
>+      fi

These four lines try to say
	[ "x$j" = "x$HOST" ] && USE_LOCAL=1

>+    done
>+    
>+    if [ "$USE_LOCAL" -eq 1 ]

You shall not compare a string to an int

>+    then
>+      URL=$i
>+    else
>+      URL="$MIRROR_SITE/`basename $i`"
>+    fi
>+  else
>+    OPTIONS="$OPTIONS $i"
>+  fi
>+done;
>+
>+$MIRROR_WGET $OPTIONS $URL
>+URL_RES=$?
>+
>+if [ "$URL_RES" = 0 ]

nah. If you're comparing an int, you're supposed to use the integer
comparison funcs.

>+then
>+  exit 0
>+fi
>+
>+# first download failed
>+if [ "$USE_LOCAL" = 1 ]

ditto.

>+then
>+  exit 1
>+fi
>+
>+if [ "$MIRROR_STRICT" = "y" ]

may barf with some test(1) impls if the var is empty. See prefixing
above

>+then
>+  exit 1;
>+fi
>+  
>+$MIRROR_WGET $OPTIONS $EXT_URL
>+
>+if [ "$?" = 0 -a "$MIRROR_GEN" = "y" ]

See last two comments.

>+then
>+  mkdir -p $MIRROR_GENDIR
>+  $MIRROR_WGET $OPTIONS -P $MIRROR_GENDIR/ $EXT_URL
>+fi