Filenames with "=" crash contrib/test_summary

Filenames with "=" crash contrib/test_summary

[Billinghurst, David (CRTS)]

The existence of file
./mips-sgi-irix6.5/mabi=64/libstdc++/testsuite/libstdc++.sum in the build
tree crashes contrib/test_summary script.

Error message is 

gawk: cmd. line:40: (FILENAME=./gcc/testsuite/g++.sum FNR=12122) fatal:
illegal name `./mips-sgi-irix6.5/mabi' in variable assignment

Crash is occurring after last line of ./gcc/testsuite/g++.sum, which is
previous file processed. gawk doesn't like doesn't like an "=" in a
filename.  I tried various quoting tricks but couldn't find a work around.

SGI awk=nawk doesn't like it either, giving

Cannot open "./mips-sgi-irix6.5/mabi=64/libstdc++/testsuite/libstdc++.sum":
No such file or directory
Input record number 26079, file
"./mips-sgi-irix6.5/mabi=64/libstdc++/testsuite/libstdc++.sum"  

+++++++++++++++++++++++++++++++++++++++++
(Mr) David Billinghurst
Comalco Research Centre
PO Box 316, Thomastown, Vic, Australia, 3074
Phone:	+61 3 9469 0642
FAX:	+61 3 9462 2700
Email:	David.Billinghurst@riotinto.com


 
>From jason@redhat.com Tue Aug 08 17:39:00 2000
From: Jason Merrill <jason@redhat.com>
To: "Martin v. Loewis" <martin@loewis.home.cs.tu-berlin.de>
Cc: gcc-bugs@gcc.gnu.org
Subject: Re: c++/8: friend template function declaration within template class
Date: Tue, 08 Aug 2000 17:39:00 -0000
Message-id: <200008090037.RAA07745@casey.cygnus.com>
X-SW-Source: 2000-08/msg00168.html
Content-length: 340

	`Jason Merrill' changed the state to `closed'.


State-Changed-From-To: analyzed-closed
State-Changed-By: jason
State-Changed-When: Tue Aug  8 17:36:46 2000
State-Changed-Why:
Fixed with
2000-08-08  Jason Merrill  <jason@redhat.com>

	* pt.c (tsubst_aggr_type): Bail if creating the argvec fails.
	(tsubst_template_arg_vector): Likewise.

Re: Filenames with "=" crash contrib/test_summary

[Alexandre Oliva]

On Aug  8, 2000, "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com> wrote:

> ./mips-sgi-irix6.5/mabi=64/libstdc++/testsuite/libstdc++.sum in the build
> tree crashes contrib/test_summary script.

> gawk: cmd. line:40: (FILENAME=./gcc/testsuite/g++.sum FNR=12122) fatal:
> illegal name `./mips-sgi-irix6.5/mabi' in variable assignment

Which version of gawk is that?  I can't reproduce this with gawk
3.0.5.

The problem seems to be that the version of gawk you're using
interprets any argument containing `=' as a variable assignment,
instead of checking whether what is on the left of the `=' sign is a
valid variable name.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

RE: Filenames with "=" crash contrib/test_summary

[Billinghurst, David (CRTS)]

Quite right.   Moving to gawk 3.0.5 (from 3.0.0) solves the problem.  Still
a portablilty issue, but probably not worth solving.

> -----Original Message-----
> From:	Alexandre Oliva [SMTP:aoliva@redhat.com]
> Sent:	Wednesday, 9 August 2000 11:14
> To:	Billinghurst, David (CRTS)
> Cc:	'gcc-bugs@gcc.gnu.org'
> Subject:	Re: Filenames with "=" crash contrib/test_summary
> 
> On Aug  8, 2000, "Billinghurst, David (CRTS)"
> <David.Billinghurst@riotinto.com> wrote:
> 
> > ./mips-sgi-irix6.5/mabi=64/libstdc++/testsuite/libstdc++.sum in the
> build
> > tree crashes contrib/test_summary script.
> 
> > gawk: cmd. line:40: (FILENAME=./gcc/testsuite/g++.sum FNR=12122) fatal:
> > illegal name `./mips-sgi-irix6.5/mabi' in variable assignment
> 
> Which version of gawk is that?  I can't reproduce this with gawk
> 3.0.5.
> 
> The problem seems to be that the version of gawk you're using
> interprets any argument containing `=' as a variable assignment,
> instead of checking whether what is on the left of the `=' sign is a
> valid variable name.
> 
> -- 
> Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
> Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
> CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
> Free Software Evangelist    *Please* write to mailing lists, not to me
>From snyder@d0sgibnl1.fnal.gov Tue Aug 08 21:46:00 2000
From: scott snyder <snyder@d0sgibnl1.fnal.gov>
To: gcc-bugs@gcc.gnu.org
Subject: g++ ICE in find_function_data
Date: Tue, 08 Aug 2000 21:46:00 -0000
Message-id: <200008090446.XAA81442@d0sgibnl1.fnal.gov>
X-SW-Source: 2000-08/msg00175.html
Content-length: 897

hi -

For a recent cvs version of gcc (2.96 20000808 on i686-pc-linux-gnu),
the compiler gives an ICE on the following input:

------------------------------------------------------------------------
struct SurfDCA
{
  SurfDCA ();
};


struct VtrackTest
{
  int* getVtrack();
  int* getVtrack(int, const SurfDCA* = new SurfDCA);
  int* getVtrack(double);
};

int* VtrackTest::getVtrack()
{
  return getVtrack(0);
}

int* VtrackTest::getVtrack(double)
{
  return getVtrack(0);
}
------------------------------------------------------------------------


$ ./cc1plus x.cc
 int *VtrackTest::getVtrack () int *VtrackTest::getVtrack () int *VtrackTest::getVtrack (double) int *VtrackTest::getVtrack (double)
x.cc:21: Internal compiler error in find_function_data, at function.c:
329
Please submit a full bug report.
See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.


thanks,
sss
>From aoliva@redhat.com Tue Aug 08 22:12:00 2000
From: Alexandre Oliva <aoliva@redhat.com>
To: "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com>
Cc: "'gcc-bugs@gcc.gnu.org'" <gcc-bugs@gcc.gnu.org>, gcc-patches@gcc.gnu.org
Subject: Re: Filenames with "=" crash contrib/test_summary
Date: Tue, 08 Aug 2000 22:12:00 -0000
Message-id: <orwvhrqaqf.fsf@guarana.lsd.ic.unicamp.br>
References: <A9E96A79C068D211A6A90000C07BDF0D88F1D9@crtsmail.crts.techaust.riotinto.com.au>
X-SW-Source: 2000-08/msg00176.html
Content-length: 562

On Aug  8, 2000, "Billinghurst, David (CRTS)" <David.Billinghurst@riotinto.com> wrote:

>> From:	Alexandre Oliva [SMTP:aoliva@redhat.com]

>> The problem seems to be that the version of gawk you're using
>> interprets any argument containing `=' as a variable assignment,
>> instead of checking whether what is on the left of the `=' sign is a
>> valid variable name.

> Quite right.  Moving to gawk 3.0.5 (from 3.0.0) solves the problem.
> Still a portablilty issue, but probably not worth solving.

It was simple enough.  I'm checking in the following patch: