1.0.3 pre-release available

1.0.3 pre-release available

[Jeffrey A Law]

I've made the egcs-1.0.3 prerelease available.

~ftp/pub/egcs/snapshots/egcs-1.0.3-prerelease

There's a full tarball and a set of diffs from 1.0.2 to the 1.0.3
prerelease.

Assuming no packaging errors are found while I'm on vacation next
week I'll tweak the version string and make the offical egcs-1.0.3
release ready when I return.

For reference, here's what's changed:

	* Fix typo in libio (HJ -- can you be more specific about
	what exactly you fixed?)

	* Fix the Fortran negative array index problem (again).

	* Fix a few Alpha code generation and compiler abort problems
	exposed by RedHat 5.1.

	* Fix 2 x86 code generation bugs exposed by glibc2 and
	RedHat 5.1.

	* Fix rs6000/ppc bug when converting from values from integer
	types to floating point types in frameless functions.

jeff

Re: 1.0.3 pre-release available

[Toon Moene]

> I've made the egcs-1.0.3 prerelease available.

> 	* Fix typo in libio (HJ -- can you be more specific about
> 	what exactly you fixed?)

> 	* Fix the Fortran negative array index problem (again).

This indeed seems to be fixed (see below).

>	* Fix a few Alpha code generation and compiler abort problems
>	exposed by RedHat 5.1.

So there were Linux/Alpha problems ?  Shees, means that Joe was  
right after all - perhaps I missed it due to the ix86 advocates  
being so much more vocal (gives a whole new meaning to "devils  
advocate" :-)

>	* Fix 2 x86 code generation bugs exposed by glibc2 and
>	RedHat 5.1.

>	* Fix rs6000/ppc bug when converting from values from integer
>	types to floating point types in frameless functions.

This is what I got on a "Red Hat Linux release 4.2 (Biltmore)"  
Alpha system (alphaev56-unknown-linux-gnulibc1 according to  
configure).

Build & compare without problems.

Result of make -k check:

                === libio tests ===


                === libio Summary ===

# of expected passes            40

                === libstdc++ tests ===

FAIL: tstring.cc compilation
FAIL: tstring.cc -O compilation

                === libstdc++ Summary ===

# of expected passes            24
# of unexpected failures        2
# of expected failures          4

                === gcc tests ===

FAIL: gcc.c-torture/execute/960218-1.c execution,  -O0
FAIL: gcc.c-torture/execute/960218-1.c execution,  -O1
FAIL: gcc.c-torture/execute/960218-1.c execution,  -O2
FAIL: gcc.c-torture/execute/960218-1.c execution,  -O2  
-fomit-frame-pointer -finline-functions
FAIL: gcc.c-torture/execute/960218-1.c execution,  -O2  
-fomit-frame-pointer -finline-functions -funroll-loops
FAIL: gcc.c-torture/execute/960218-1.c execution,  -O2  
-fomit-frame-pointer -finline-functions -funroll-all-loops
FAIL: gcc.c-torture/execute/complex-5.c execution,  -O0
FAIL: gcc.c-torture/execute/complex-5.c execution,  -O1
FAIL: gcc.c-torture/execute/complex-5.c execution,  -O2
FAIL: gcc.c-torture/execute/complex-5.c execution,  -O2  
-fomit-frame-pointer -finline-functions
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution,  -O0
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution,  -O1
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution,  -O2
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution,  -O2  
-fomit-frame-pointer -finline-functions

                === gcc Summary ===

# of expected passes            4856
# of unexpected failures        14
# of expected failures          5
# of unsupported tests          9

                === g++ tests ===

FAIL: g++.benjamin/p12475.C  (test for warnings, line 4)
XPASS: g++.brendan/array1.C overflow in array dimension.* , (test  
for errors, line 6)
XPASS: g++.jason/destruct3.C - (test for bogus messages, line 38)
XPASS: g++.mike/dyncast1.C  Execution test
XPASS: g++.mike/dyncast2.C  Execution test
FAIL: g++.mike/eh39.C  Execution test
FAIL: g++.mike/eh40.C  Execution test

                === g++ Summary ===

# of expected passes            3388
# of unexpected failures        3
# of unexpected successes       4
# of expected failures          82
# of untested testcases         10

                === g77 tests ===

FAIL: g77.f-torture/execute/dnrm2.f execution,  -O0
FAIL: g77.f-torture/execute/dnrm2.f execution,  -O1

                === g77 Summary ===

# of expected passes            130
# of unexpected failures        2

Furthermore, I tried the following Fortran example, which fails  
with egcs-1.0.2:

      program rotor
      implicit none
      integer   KaKc(0:5,-5:5,0:2)
      integer   cj
      call setkac( KaKc, 5 )
      stop
      end
c     ---------------------------------------------
      subroutine setkac( KaKc, Jmax )
      implicit none
      integer   Jmax
      integer   KaKc( 0:Jmax, -Jmax:Jmax, 0:2 )
      integer   cj, i
      cj = 0
      i = -cj
      KaKc( cj, i, 1 ) = 0
      write(*,*) 'Here we go !'
      KaKc( cj, -cj, 1 ) = 0
      return
      end

which would bomb after the message "Here we go !", and now  
completes succesfully.

Cheers,
Toon.

Re: 1.0.3 pre-release available

[Jeffrey A Law]

  In message < m0yTUQG-000598C@ocean.lucon.org >you write:
  > Ok. I will give it one more try. Please let me know what else I
  > can do to explain it clearer.
Sorry, I didn't mean to imply you hadn't provided this information
already -- I knew you had, but I couldn't easily find it in my
saved messages.

Thanks for the URLs, they have the information I need.


jeff

Re: 1.0.3 pre-release available

[Jeffrey A Law]

  In message < 9804261449.AA17362@moene.indiv.nluug.nl >you write:
  > So there were Linux/Alpha problems ?  Shees, means that Joe was  
  > right after all - perhaps I missed it due to the ix86 advocates  
  > being so much more vocal (gives a whole new meaning to "devils  
  > advocate" :-)
There was a profiling problem and problems compiling ImageMagick
and emacs-20.2.  I've expanded on the "Alpha bugs" section in the
propsed release announcement :-)


  >                 === libstdc++ tests ===
  > 
  > FAIL: tstring.cc compilation
  > FAIL: tstring.cc -O compilation
Hmmm, this worries me a little.  Then again, maybe this has been
failing all along.


  > FAIL: gcc.c-torture/execute/960218-1.c execution,  -O0
The "glob" problem with the Linux C library.

  > FAIL: gcc.c-torture/execute/complex-5.c execution,  -O0
The usual complex failures.  

  > FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution,  -O0
Not sure.  This might be another case where the alpha needs
-mieee.  Dunno.

jeff

Re: 1.0.3 pre-release available

[H.J. Lu]

> 
> For reference, here's what's changed:
> 
> 	* Fix typo in libio (HJ -- can you be more specific about
> 	what exactly you fixed?)
> 

Ok. I will give it one more try. Please let me know what else I
can do to explain it clearer.


H.J.
--

Re: 1.0.3 pre-release available

[Michael Alan Dorman]

Jeffrey A Law <law@cygnus.com> writes:
> There was a profiling problem and problems compiling ImageMagick
> and emacs-20.2.  I've expanded on the "Alpha bugs" section in the
> propsed release announcement :-)

Ah, I just ran into the emacs issue yesterday.  It's great that it's
fixed.  Our normal gcc-compiling-person will hopefully have the
pre-release pacakged soon.

>   > FAIL: tstring.cc compilation
>   > FAIL: tstring.cc -O compilation
> Hmmm, this worries me a little.  Then again, maybe this has been
> failing all along.

Is there anyone that you know of consistently running the test suite
against the compiler on Linux/Alpha?  If not, I'd certainly volunteer
to do this, if it's useful.  I guess I kind of figured that with RTH
here, it was probably covered.

Mike.

Re: 1.0.3 pre-release available

[Nathan Myers]

H.J. Lu wrote:
> 
> >
> > For reference, here's what's changed:
> >
> >       * Fix typo in libio (HJ -- can you be more specific about
> >       what exactly you fixed?)
> >
> 
> Ok. I will give it one more try. Please let me know what else I
> can do to explain it clearer.

For what it's worth, this patch has also been applied to the unsupported/isolib
source.  

Nathan Myers
ncm@cantrip.org

Re: 1.0.3 pre-release available

[Richard Henderson]

On Sun, Apr 26, 1998 at 09:03:34PM -0400, Michael Alan Dorman wrote:
> Ah, I just ran into the emacs issue yesterday.  It's great that it's
> fixed.  Our normal gcc-compiling-person will hopefully have the
> pre-release pacakged soon.

Do make sure emacs is fixed for you.  Cristian Gafton had reported
a problem, but not given details.  Later he reported that he could
not reproduce it.  It could have been the same problem as afflicted
ImageMagik, which would explain things, but it would be good to 
have this verified.

> Is there anyone that you know of consistently running the test suite
> against the compiler on Linux/Alpha?

Alexandre Oliva runs the tests regularly on a whole suite of systems,
including alpha-unknown-linux-gnu.  I run them myself often, though I
don't always do anything about failures immediately.  ;-)

I don't recall what the tstring.cc failure is.  I'm much worse about
following up on c++ failures vs c/f77 failures.

The only two I recall that I expect to fail are complex-5 and fp-cmp-1.


r~

Re: 1.0.3 pre-release available

[H.J. Lu]

> 
> I don't recall what the tstring.cc failure is.  I'm much worse about
> following up on c++ failures vs c/f77 failures.
> 

I think that may be caused by the new binutils and old glibc. At
least, that was what I saw on alpha. Please make sure you follow
my release note if you use binutils newer than 2.8.1. I don't think
the official binutils mentions it at all.

-- 
H.J. Lu (hjl@gnu.org)