Does anyone check libstdc++-v3 failures in gcc 3.0?

Does anyone check libstdc++-v3 failures in gcc 3.0?

[H . J . Lu]

Does anyone check those libstdc++v3 failures in gcc 3.0? I saw quite
a few failures on Linux/ia32/glibc 2.2:

+: pass, -b: build failure, -r: run failure, x: disabled
------------------------------------------------------------------------
static  shared  test
------------------------------------------------------------------------
-r      +       21_strings/ctor_copy_dtor.cc
-r      +       21_strings/element_access.cc
-r      +       21_strings/insert.cc
-r      +       21_strings/substr.cc
-b      -b      23_containers/map_operators.cc
-b      -b      23_containers/set_operators.cc
-b      -b      26_numerics/c99_macros.cc
-b      -b      27_io/filebuf.cc
-r      -r      27_io/ios_base_members_static.cc
-r      -r      27_io/istream_extractor_char.cc
-b      -b      27_io/istream_seeks.cc
+       +       27_io/istream_sentry.cc
diff: ./istream_unformatted-2.txt: No such file or directory
+       diff: ./istream_unformatted-2.txt: No such file or directory
+       27_io/istream_unformatted.cc

I don't think it is too hard to modify contrib/test_summary and
libstdc++-v3/mkcheck.in to send in test summary for libstdc++-v3.
It will help keeping track the status of libstdc++-v3. I can provide
a patch if everyone agrees it is a good idea and noone has any time.
I hope gcc 3.0 will ship a working libstdc++ which at least passes its
own testsuite. That is what gcc had been doing before gcc 3.0.

Thanks.


H.J.

Re: Does anyone check libstdc++-v3 failures in gcc 3.0?

[Phil Edwards]

On Sun, Apr 15, 2001 at 08:49:49PM -0700, H . J . Lu wrote:
> Does anyone check those libstdc++v3 failures in gcc 3.0?

Every day.


> -b      -b      23_containers/map_operators.cc
> -b      -b      23_containers/set_operators.cc

XFAILs.  The mkcheck script doesn't know about XFAIL yet.

> -b      -b      26_numerics/c99_macros.cc

glibc defines certain things as macros which C++ requires to be functions.
I don't recall whether this is supposed to be XFAIL or not (I think it
should pass).

> -b      -b      27_io/filebuf.cc

Undefined references at link time.  Looks like 'short' is being used as
the instantiating character type (where did wchar_t go).  I haven't looked
into this one.  It's working on the trunk.

> -b      -b      27_io/istream_seeks.cc

The compiler spits out a bunch of "by the way, I'm choosing this conversion
over that conversion because it's a better match" messages, and then
simply /stops/.  Apparently those messages are supposed to be errors.
(This was discussed once before when it was fixed on the trunk; I'm sorry
but I still don't understand why these errors sound more like "informational"
or "debug" messsages.)  Fixed on the trunk.

> diff: ./istream_unformatted-2.txt: No such file or directory
> +       diff: ./istream_unformatted-2.txt: No such file or directory

Missing empty file on the branch.  Fixed on the trunk.


> I hope gcc 3.0 will ship a working libstdc++ which at least passes its
> own testsuite. That is what gcc had been doing before gcc 3.0.

The trunk shows fewer failures.


-- 
pedwards at disaster dot jaj dot com  |  pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools.  Fools are protected by more capable fools.

Re: Does anyone check libstdc++-v3 failures in gcc 3.0?

[H . J . Lu]

On Mon, Apr 16, 2001 at 12:14:49AM -0400, Phil Edwards wrote:
> 
> > I hope gcc 3.0 will ship a working libstdc++ which at least passes its
> > own testsuite. That is what gcc had been doing before gcc 3.0.
> 
> The trunk shows fewer failures.

Any particular reason not to fix the gcc 3.0 branch?



H.J.

Re: Does anyone check libstdc++-v3 failures in gcc 3.0?

[Gabriel Dos Reis]

Phil Edwards <pedwards@disaster.jaj.com> writes:

| On Sun, Apr 15, 2001 at 08:49:49PM -0700, H . J . Lu wrote:
| > Does anyone check those libstdc++v3 failures in gcc 3.0?
| 
| Every day.
| 
| 
| > -b      -b      23_containers/map_operators.cc
| > -b      -b      23_containers/set_operators.cc
| 
| XFAILs.  The mkcheck script doesn't know about XFAIL yet.

But the new testing framework does know.

-- Gaby

Re: Does anyone check libstdc++-v3 failures in gcc 3.0?

[Phil Edwards]

On Mon, Apr 16, 2001 at 09:17:48AM +0200, Gabriel Dos Reis wrote:
> Phil Edwards <pedwards@disaster.jaj.com> writes:
> | 
> | XFAILs.  The mkcheck script doesn't know about XFAIL yet.
> 
> But the new testing framework does know.

Yes.  One more reason in favor of merging trunk to branch for the library.


Phil

-- 
pedwards at disaster dot jaj dot com  |  pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools.  Fools are protected by more capable fools.