Patchwork [libstdc++,google] Make bits/stamp-bits generated by libstc++-v3/include a regular file

login
register
mail settings
Submitter Simon Baldwin
Date Feb. 7, 2011, 4:34 p.m.
Message ID <20110207163412.DCDD71C6839@hpgntab-ubiq73.eem.corp.google.com>
Download mbox | patch
Permalink /patch/82117/
State New
Headers show

Comments

Simon Baldwin - Feb. 7, 2011, 4:34 p.m.
Make bits/stamp-bits generated by libstc++-v3/include a regular file.

Building stamp-bits-sup in libstdc++-v3/include creates a bits/stamp-bits
file.  Because it's generated by 'ln -s stamp-bits .', the file is a
symbolic link that points back to itself.  This prevents cpio -L from
copying the built tree ("Too many levels of symbolic links").

This change removes the self-referential bits/stamp-bits and replaces it
with a regular timestamp file, allowing cpio -L to handle it cleanly.

Primarily needed for google/integration.  OK?

Secondarily, equally applicable to trunk.  OK there also?

libstdc++-v3/ChangeLog:
2011-02-07  Simon Baldwin  <simonb@google.com>

	* include/Makefile.am: Fix links built for stamp-sup-bits.
	* include/Makefile.in: Rebuild.
Diego Novillo - Feb. 7, 2011, 5:07 p.m.
On Mon, Feb 7, 2011 at 11:34, Simon Baldwin <simonb@google.com> wrote:

> This change removes the self-referential bits/stamp-bits and replaces it
> with a regular timestamp file, allowing cpio -L to handle it cleanly.
>
> Primarily needed for google/integration.  OK?

Yes.

> Secondarily, equally applicable to trunk.  OK there also?

I'll defer to the libstdc++ folks for this one.


Diego.
Paolo Carlini - Feb. 7, 2011, 5:22 p.m.
On 02/07/2011 06:07 PM, Diego Novillo wrote:
> I'll defer to the libstdc++ folks for this one.
>   
In turn, before approving it, in particular such late in the 4.6.0
release process, I'd like to ear from Ralf...

Thanks,
Paolo.
Ralf Wildenhues - Feb. 7, 2011, 6:49 p.m.
* Paolo Carlini wrote on Mon, Feb 07, 2011 at 06:22:48PM CET:
> On 02/07/2011 06:07 PM, Diego Novillo wrote:
> > I'll defer to the libstdc++ folks for this one.
> >   
> In turn, before approving it, in particular such late in the 4.6.0
> release process, I'd like to ear from Ralf...

Looks safe enough to me.

Eventually, this whole stamp file stuff in v3/include should probably
go, it is fragile and error-prone, and probably not needed.  4.7
material.

Thanks,
Ralf
Paolo Carlini - Feb. 7, 2011, 6:53 p.m.
Hi,
>>> I'll defer to the libstdc++ folks for this one.
>>>   
>>>       
>> In turn, before approving it, in particular such late in the 4.6.0
>> release process, I'd like to ear from Ralf...
>>     
> Looks safe enough to me.
>   
Ok then.
> Eventually, this whole stamp file stuff in v3/include should probably
> go, it is fragile and error-prone, and probably not needed.  4.7
> material.
>   
I see, thanks for the suggestion, let's remember it when 4.6 branches!

Paolo.

Patch

Index: libstdc++-v3/include/Makefile.am
===================================================================
--- libstdc++-v3/include/Makefile.am	(revision 169786)
+++ libstdc++-v3/include/Makefile.am	(working copy)
@@ -928,6 +928,7 @@  stamp-bits: ${bits_headers}
 
 stamp-bits-sup: stamp-bits ${bits_sup_headers}
 	@-cd ${bits_builddir} && $(LN_S) $? . 2>/dev/null
+	@-cd ${bits_builddir} && rm -f stamp-bits && $(STAMP) stamp-bits
 	@$(STAMP) stamp-bits-sup
 
 stamp-c_base: ${c_base_headers}
Index: libstdc++-v3/include/Makefile.in
===================================================================
--- libstdc++-v3/include/Makefile.in	(revision 169786)
+++ libstdc++-v3/include/Makefile.in	(working copy)
@@ -1326,6 +1326,7 @@  stamp-bits: ${bits_headers}
 
 stamp-bits-sup: stamp-bits ${bits_sup_headers}
 	@-cd ${bits_builddir} && $(LN_S) $? . 2>/dev/null
+	@-cd ${bits_builddir} && rm -f stamp-bits && $(STAMP) stamp-bits
 	@$(STAMP) stamp-bits-sup
 
 stamp-c_base: ${c_base_headers}