Thread: regression test failure

regression test failure

From
Neil Conway
Date:
When I run the regression tests against current sources, I get
failures because bison-generated error messages use "parse
error", not "syntax error". I vaguely recall running into
this issue before I left for the summer -- did we resolve
it?

[neil@arch neil]$ uname -a
FreeBSD arch.wavefire.com 5.1-CURRENT FreeBSD 5.1-CURRENT #9: Fri Jul 25 12:10:58 PDT 2003
root@arch.wavefire.com:/usr/obj/usr/src/sys/WORKSTATION-5.0-SMP i386 
[neil@arch neil]$ bison --version
bison (GNU Bison) 1.875
[neil@arch neil]$ flex --version
flex version 2.5.4

regression.diffs is attached.

-Neil

P.S. Someone should go ahead and re-generate INSTALL from
the latest SGML: INSTALL still claims that we need bison
1.50 or greater.


Attachment

Re: regression test failure

From
Bruce Momjian
Date:
Strange.  I know we check for bison >= 1.875, and you have that, and so
do I, but I don't see those regression failures.  Is it possible you
have old bison output files from an older bison release?

---------------------------------------------------------------------------

Neil Conway wrote:
> When I run the regression tests against current sources, I get
> failures because bison-generated error messages use "parse
> error", not "syntax error". I vaguely recall running into
> this issue before I left for the summer -- did we resolve
> it?
> 
> [neil@arch neil]$ uname -a
> FreeBSD arch.wavefire.com 5.1-CURRENT FreeBSD 5.1-CURRENT #9: Fri Jul 25 12:10:58 PDT 2003
root@arch.wavefire.com:/usr/obj/usr/src/sys/WORKSTATION-5.0-SMP i386
 
> [neil@arch neil]$ bison --version
> bison (GNU Bison) 1.875
> [neil@arch neil]$ flex --version
> flex version 2.5.4
> 
> regression.diffs is attached.
> 
> -Neil
> 
> P.S. Someone should go ahead and re-generate INSTALL from 
> the latest SGML: INSTALL still claims that we need bison
> 1.50 or greater.
> 

[ Attachment, skipping... ]

> 
> ---------------------------(end of broadcast)---------------------------
> TIP 3: if posting/reading through Usenet, please send an appropriate
>       subscribe-nomail command to majordomo@postgresql.org so that your
>       message can get through to the mailing list cleanly

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
359-1001+  If your life is a hard drive,     |  13 Roberts Road +  Christ can be your backup.        |  Newtown Square,
Pennsylvania19073
 


Re: regression test failure

From
Tom Lane
Date:
Bruce Momjian <pgman@candle.pha.pa.us> writes:
> Strange.  I know we check for bison >= 1.875, and you have that, and so
> do I, but I don't see those regression failures.  Is it possible you
> have old bison output files from an older bison release?

I think the check is only a warning though; and the only thing that
actually fails to build is ecpg's preproc.y.  It's possible his current
copy of parser/gram.c was built with an older bison before he hit the
hard failure, and then he didn't go back and rebuild after updating
bison.

I'd suggest a "make maintainer-clean" and try again.
        regards, tom lane


Re: regression test failure

From
Neil Conway
Date:
On Thu, Aug 07, 2003 at 04:09:33PM -0400, Tom Lane wrote:I think the check is only a warning though; and the only thing
that
> actually fails to build is ecpg's preproc.y.  It's possible his current
> copy of parser/gram.c was built with an older bison before he hit the
> hard failure, and then he didn't go back and rebuild after updating
> bison.

Uh, yeah -- of course, that's exactly what the problem was. I had
assumed, wrongly of course, that "make distclean" removed bison
output.

I feel stupid :-)

-Neil



Re: regression test failure

From
Bruce Momjian
Date:
I didn't know about "make maintainer-clean" either.

---------------------------------------------------------------------------

Neil Conway wrote:
> On Thu, Aug 07, 2003 at 04:09:33PM -0400, Tom Lane wrote:
>  I think the check is only a warning though; and the only thing that
> > actually fails to build is ecpg's preproc.y.  It's possible his current
> > copy of parser/gram.c was built with an older bison before he hit the
> > hard failure, and then he didn't go back and rebuild after updating
> > bison.
> 
> Uh, yeah -- of course, that's exactly what the problem was. I had
> assumed, wrongly of course, that "make distclean" removed bison
> output.
> 
> I feel stupid :-)
> 
> -Neil
> 
> 

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
359-1001+  If your life is a hard drive,     |  13 Roberts Road +  Christ can be your backup.        |  Newtown Square,
Pennsylvania19073