On Tuesday 13 January 2009, Andrew Stubbs wrote:
I think this patch should work for you. It works for /bin/bash and I
also tried a few bits and pieces from /usr/bin/X11.
Yes, with this version I get a correct build of the installer. Thanks.
If I build the amd64 cdrom target of D-I, I get a
On Sat, Jan 17, 2009 at 11:18:19AM +0100, Frans Pop wrote:
Bastian: any objections to committing the last version of the patch in
this BR?
Will take care of it in the next days.
Bastian
--
Without facts, the decision cannot be made logically. You must rely on
your human intuition.
Frans Pop wrote:
To reproduce (assuming you have debian sid or lenny installed) you'll
need to do something like this (besides building and installing your
version of mklibs and mklibs-copy of course):
$ apt-get install build-essential subversion
$ svn co
Frans Pop wrote:
That still gives me an error:
Traceback (most recent call last):
File /usr/bin/mklibs, line 509, in module
library_symbols_used[lib].add(library_symbols[lib][name])
KeyError: 'strlong...@base@unknown'
make[2]: *** [stamps/tree-netboot-stamp] Error 1
OK, let's try again.
On Monday 12 January 2009, Andrew Stubbs wrote:
Frans Pop wrote:
That still gives me an error:
Traceback (most recent call last):
File /usr/bin/mklibs, line 509, in module
library_symbols_used[lib].add(library_symbols[lib][name])
KeyError: 'strlong...@base@unknown'
make[2]: ***
Frans Pop wrote:
I've just given your patch a try, but if I try to build the Debian
Installer (which in Debian is the main user of mklibs) with it on a
system running Debian/unstable amd64, I get the following error:
Sorry for the delay, I've been busy.
Please try the attached updated patch.
On Thursday 08 January 2009, Andrew Stubbs wrote:
Sorry for the delay, I've been busy.
No problem. Thanks for the updated patch.
Please try the attached updated patch. I've added a fall back for
@unknown symbols. This is basically the same as the previous
implementation - it just selects the
Hello Andrew,
Thank you for this proposed patch. Although I'm far from an expert on
this, your approach looks logical.
BTW, sorry for the late response. Main reason is that development was
somewhat frozen due to the preparations for Lenny.
This patch modifies mklibs-readelf to extract the
Package: mklibs
Tags: patch
The problem is that mklibs selects either the first symbol of the
correct name, or the one in libc, if there is one. It completely
disregards the information left by the linker defining where each symbol
*should* come from.
The dynamic linker does use this data,
9 matches
Mail list logo
