This case was originally submitted tot he binutils project by Richard Stallman.
> I think this is a matter of incorrect output from ld, because the
> second error message doesn't follow the specs in the GNU Coding
> Standards. Outputting that extra information (in parens) may
> be useful, but please don't put it into the beginning of the error
> message which is the part that's supposed to be parsable by programs.
This was in reference to an emacs bug report submitted by firstname.lastname@example.org where
was talking about parsing this error message from the linker:
> sip/sip.a(Receiver.o)(.text+0x1ec0): In function `Receiver::ProcessMsg(int)':
> /opt/users/ceder/sipfw/sip/src/Receiver.cpp:776: undefined reference to
> `scAccept(int, int)'
undefined reference to `scAccept(int, int)'
Created attachment 318 [details]
Testcase with makefile and source files
A small testcase to reproduce the incorrect error messages
I have created a testcase to reproduce this problem
Created attachment 319 [details]
Possible fix for PR614 - regression testing currently in progress
Oops - the testcase is broken. Please replace "-L" with "-L." in the last line
of the "doit" target.
Regression checking complete - no regressions.
No waiting on RMS's approval for the output format before committing.
2005-01-05 Nick Clifton <email@example.com>
* ldmisc.c (vfinfo): Alter output to conform to the GNU Coding
Standard's specification for parsable error messages.