You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@stdcxx.apache.org by Geoffrey Winn <ge...@googlemail.com> on 2006/09/22 16:34:20 UTC
Problem running stdcxx tests on Linux
I've built optimised and debug versions of stdcxx on Linux and I've built
the tests as described in Section 7 of the README. However, I'm having
trouble working out whether they worked or not. There were no obvious errors
before I did the gmake run to execute the tests. The run produces the
results below. (I've attached a file with the results in as well in case
that preserves the formatting a little better.)
My main question is what happened? It looks like all the tests failed, but
there are no error messages. Is there a log file somewhere that I could look
at?
Thanks in advance,
Geoff.
--- NAME STATUS ASSRTS HIGH DELTA FAILED LOW RGRESS
PERCNT
1 0.printf FORMAT 0 0 0 0 0 0
0%
2 18.limits.traps 1 0 0 0 0 0 0
0%
3 19.cerrno 1 0 0 0 0 0 0
0%
4 2.smartptrared 1 0 0 0 0 0 0
0%
5 2.smartptr.weak 1 0 0 0 0 0 0
0%
6 21.string.assign 1 0 0 0 0 0 0
0%
7 21.string.insert 1 0 0 0 0 0 0
0%
8 22.locale.money.ge 1 0 0 0 0 0 0
0%
9 22.locale.money.pu 1 0 0 0 0 0 0
0%
10 22.locale.num.get 1 0 0 0 0 0 0
0%
11 22.locale.num.put 1 0 0 0 0 0 0
0%
12 25.adjacent.find 1 0 0 0 0 0 0
0%
13 25.copy 1 0 0 0 0 0 0
0%
14 25.equal 1 0 0 0 0 0 0
0%
15 25.fill 1 0 0 0 0 0 0
0%
16 25.find.end 1 0 0 0 0 0 0
0%
17 25.find.first 1 0 0 0 0 0 0
0%
18 25.find 1 0 0 0 0 0 0
0%
19 25.for.each 1 0 0 0 0 0 0
0%
20 25.generate 1 0 0 0 0 0 0
0%
21 25.mismatch 1 0 0 0 0 0 0
0%
22 26.c.math 1 0 0 0 0 0 0
0%
--- 1 0 0 0 0 0 0
0%
FAIL COMP LINK ASSERT CORE KILL RGRESS
21 0 0 0 0 0 0
Re: Problem running stdcxx tests on Linux
Posted by Martin Sebor <se...@roguewave.com>.
Geoffrey Winn wrote:
> Thank you both for that quick reply. I'm really trying to establish that I
> have a sound stdcxx build, installation and so forth. I'm not worried about
> some bugs in the code - especially if you already know about them :-)
>
> What concerned me in this case was the appearance that _all_ the tests had
> failed totally which suggested that something much more basic had gone
> wrong
> eg I had never built the library correctly in the first place.
I understand, it would worry me too. I would suggest giving
the examples a try instead of the tests, they should all be
fine and should give you at least some degree of confidence
in the build. Again, feel free to ask if you see any issues.
Martin
>
> I'll move on to the Linux version of SDO now.
>
> Thanks again.
>
> Geoff.
>
> On 22/09/06, Martin Sebor <se...@roguewave.com> wrote:
>
>>
>> Geoffrey Winn wrote:
>> > I've built optimised and debug versions of stdcxx on Linux and I've
>> > built the tests as described in Section 7 of the README. However, I'm
>> > having trouble working out whether they worked or not. There were no
>> > obvious errors before I did the gmake run to execute the tests. The run
>> > produces the results below. (I've attached a file with the results
>> in as
>> > well in case that preserves the formatting a little better.)
>> >
>> > My main question is what happened? It looks like all the tests failed,
>> > but there are no error messages. Is there a log file somewhere that I
>> > could look at?
>>
>> The problem is that the test driver is incompatible with the test
>> harness in that release (the harness passes in the wrong options),
>> and the tests fail with the status of 1 because of it. You can run
>> each test by hand instead, or you can apply the following patch., it
>> should fix it: http://svn.apache.org/viewvc?view=rev&revision=419995
>>
>> Btw., I think I might have already mentioned in a previous post,
>> the stdcxx test suite was (and still is) being migrated from the
>> Rogue Wave test driver and harness and not all tests are fully
>> stable on all platforms (although gcc/Linux should be pretty clean).
>> Failures may indicate known problems in the library (for which we
>> have issues in Jira), problems in the tests themselves, or even
>> in the compiler or the OS. If you're concerned about any test or
>> failure in particular let us know, we'll be happy to explain it.
>>
>> Martin
>>
>
Re: Problem running stdcxx tests on Linux
Posted by Geoffrey Winn <ge...@googlemail.com>.
Thank you both for that quick reply. I'm really trying to establish that I
have a sound stdcxx build, installation and so forth. I'm not worried about
some bugs in the code - especially if you already know about them :-)
What concerned me in this case was the appearance that _all_ the tests had
failed totally which suggested that something much more basic had gone wrong
eg I had never built the library correctly in the first place.
I'll move on to the Linux version of SDO now.
Thanks again.
Geoff.
On 22/09/06, Martin Sebor <se...@roguewave.com> wrote:
>
> Geoffrey Winn wrote:
> > I've built optimised and debug versions of stdcxx on Linux and I've
> > built the tests as described in Section 7 of the README. However, I'm
> > having trouble working out whether they worked or not. There were no
> > obvious errors before I did the gmake run to execute the tests. The run
> > produces the results below. (I've attached a file with the results in as
> > well in case that preserves the formatting a little better.)
> >
> > My main question is what happened? It looks like all the tests failed,
> > but there are no error messages. Is there a log file somewhere that I
> > could look at?
>
> The problem is that the test driver is incompatible with the test
> harness in that release (the harness passes in the wrong options),
> and the tests fail with the status of 1 because of it. You can run
> each test by hand instead, or you can apply the following patch., it
> should fix it: http://svn.apache.org/viewvc?view=rev&revision=419995
>
> Btw., I think I might have already mentioned in a previous post,
> the stdcxx test suite was (and still is) being migrated from the
> Rogue Wave test driver and harness and not all tests are fully
> stable on all platforms (although gcc/Linux should be pretty clean).
> Failures may indicate known problems in the library (for which we
> have issues in Jira), problems in the tests themselves, or even
> in the compiler or the OS. If you're concerned about any test or
> failure in particular let us know, we'll be happy to explain it.
>
> Martin
>
Re: Problem running stdcxx tests on Linux
Posted by Martin Sebor <se...@roguewave.com>.
Geoffrey Winn wrote:
> I've built optimised and debug versions of stdcxx on Linux and I've
> built the tests as described in Section 7 of the README. However, I'm
> having trouble working out whether they worked or not. There were no
> obvious errors before I did the gmake run to execute the tests. The run
> produces the results below. (I've attached a file with the results in as
> well in case that preserves the formatting a little better.)
>
> My main question is what happened? It looks like all the tests failed,
> but there are no error messages. Is there a log file somewhere that I
> could look at?
The problem is that the test driver is incompatible with the test
harness in that release (the harness passes in the wrong options),
and the tests fail with the status of 1 because of it. You can run
each test by hand instead, or you can apply the following patch., it
should fix it: http://svn.apache.org/viewvc?view=rev&revision=419995
Btw., I think I might have already mentioned in a previous post,
the stdcxx test suite was (and still is) being migrated from the
Rogue Wave test driver and harness and not all tests are fully
stable on all platforms (although gcc/Linux should be pretty clean).
Failures may indicate known problems in the library (for which we
have issues in Jira), problems in the tests themselves, or even
in the compiler or the OS. If you're concerned about any test or
failure in particular let us know, we'll be happy to explain it.
Martin
Re: Problem running stdcxx tests on Linux
Posted by Andrew Black <ab...@roguewave.com>.
Greetings Geoffrey.
Test output is stored in a file named like '<testname>.out', residing in
the tests subdirectory of your build directory, along side the test.
For example, the output for the '18.limits.traps' test would be saved as
'18.limits.traps.out'.
The 'STATUS' column in the output indicates the exit code that the test
terminated with. In this case, the tests exited with a status code of 1.
I believe the cause of the behavior you are experiencing is a bug in the
makefile infrastructure (STDCXX-254). This bug has been fixed in
subversion. You should be able to pull the patch from JIRA (
http://issues.apache.org/jira/browse/STDCXX-254 ) and apply it to your
source directory.
--Andrew Black
Geoffrey Winn wrote:
> I've built optimised and debug versions of stdcxx on Linux and I've built
> the tests as described in Section 7 of the README. However, I'm having
> trouble working out whether they worked or not. There were no obvious
> errors
> before I did the gmake run to execute the tests. The run produces the
> results below. (I've attached a file with the results in as well in case
> that preserves the formatting a little better.)
>
> My main question is what happened? It looks like all the tests failed, but
> there are no error messages. Is there a log file somewhere that I could
> look
> at?
>
> Thanks in advance,
>
> Geoff.
>
> --- NAME STATUS ASSRTS HIGH DELTA FAILED LOW RGRESS
> PERCNT
>
> 1 0.printf FORMAT 0 0 0 0 0 0
> 0%
>
> 2 18.limits.traps 1 0 0 0 0 0 0
> 0%
>
> 3 19.cerrno 1 0 0 0 0 0 0
> 0%
>
> 4 2.smartptrared 1 0 0 0 0 0 0
> 0%
>
> 5 2.smartptr.weak 1 0 0 0 0 0 0
> 0%
>
> 6 21.string.assign 1 0 0 0 0 0 0
> 0%
>
> 7 21.string.insert 1 0 0 0 0 0 0
> 0%
>
> 8 22.locale.money.ge 1 0 0 0 0 0 0
> 0%
>
> 9 22.locale.money.pu 1 0 0 0 0 0 0
> 0%
>
> 10 22.locale.num.get 1 0 0 0 0 0 0
> 0%
>
> 11 22.locale.num.put 1 0 0 0 0 0 0
> 0%
>
> 12 25.adjacent.find 1 0 0 0 0 0 0
> 0%
>
> 13 25.copy 1 0 0 0 0 0 0
> 0%
>
> 14 25.equal 1 0 0 0 0 0 0
> 0%
>
> 15 25.fill 1 0 0 0 0 0 0
> 0%
>
> 16 25.find.end 1 0 0 0 0 0 0
> 0%
>
> 17 25.find.first 1 0 0 0 0 0 0
> 0%
>
> 18 25.find 1 0 0 0 0 0 0
> 0%
>
> 19 25.for.each 1 0 0 0 0 0 0
> 0%
>
> 20 25.generate 1 0 0 0 0 0 0
> 0%
>
> 21 25.mismatch 1 0 0 0 0 0 0
> 0%
>
> 22 26.c.math 1 0 0 0 0 0 0
> 0%
>
>
>
> --- 1 0 0 0 0 0 0
> 0%
>
>
>
> FAIL COMP LINK ASSERT CORE KILL RGRESS
>
> 21 0 0 0 0 0 0
>