You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stdcxx.apache.org by Martin Sebor <se...@roguewave.com> on 2006/12/19 23:04:50 UTC
failed examples reported as passing with MSVC or vice versa
While going through our MSVC 8.0 jumbled up log I came across
what looks like incorrect behavior of either the exec utility
or the Windows infrastructure. What seems to be happening is
that while exec reports some examples as passing the rest of
the infrastructure actually says they failed. Below are lines
from the log relevant to the accumulate.exe example.
Andrew and/or Farid, could you guys look into it to see which
part of the infrastructure is broken?
Thanks
Martin
422>Performing Custom Build Step
422>running accumulate.exe
422>NAME STATUS WARN ASSERTS FAILED PERCNT USER
SYS REAL
422>accumulate.exe 0 0
0.330
422>example output differs from the expected
422>example failed! Exit code 0
422>
Re: failed examples reported as passing with MSVC or vice versa
Posted by Andrew Black <ab...@roguewave.com>.
Martin Sebor wrote:
> Andrew Black wrote:
>
> [...]
>> My reading of this message is that the runall.wsf script failed to
>> complete a file system command. Unfortunately, I am having
>> difficulties determining where the failure occurred, as the line
>> number is meaningless within the context of a text editor. I tend to
>> suspect that this translates to line 118 of the runall.wsf script. A
>> potential contributing cause is the permissions of the build directory
>> used by the nightly testing system.
>
> Permissions on Windows? That's something new! Could you look
> into this theory some more when you have a spare moment? We
> need to get those builds working at some point in the near
> future.
Greetings Martin.
I performed a manual (15d) build on our shared Windows 2003 AMD64 box,
and I was able to replicate the behavior using the glue script used in
the nightly build process. However, this behavior wasn't replicated
when I built the .examples project through the devenv GUI. I need to
try building the solution using the build_msvc-8.0-x64.bat script, but I
am currently unable to due to fallout from some internal infrastructure
changes.
--Andrew Black
>
> Thanks!
> Martin
Re: failed examples reported as passing with MSVC or vice versa
Posted by Martin Sebor <se...@roguewave.com>.
Andrew Black wrote:
[...]
> My reading of this message is that the runall.wsf script failed to
> complete a file system command. Unfortunately, I am having difficulties
> determining where the failure occurred, as the line number is
> meaningless within the context of a text editor. I tend to suspect that
> this translates to line 118 of the runall.wsf script. A potential
> contributing cause is the permissions of the build directory used by the
> nightly testing system.
Permissions on Windows? That's something new! Could you look
into this theory some more when you have a spare moment? We
need to get those builds working at some point in the near
future.
Thanks!
Martin
>
> --Andrew Black
>
> Martin Sebor wrote:
>
>> While going through our MSVC 8.0 jumbled up log I came across
>> what looks like incorrect behavior of either the exec utility
>> or the Windows infrastructure. What seems to be happening is
>> that while exec reports some examples as passing the rest of
>> the infrastructure actually says they failed. Below are lines
>> from the log relevant to the accumulate.exe example.
>>
>> Andrew and/or Farid, could you guys look into it to see which
>> part of the infrastructure is broken?
>>
>> Thanks
>> Martin
>>
>> 422>Performing Custom Build Step
>> 422>running accumulate.exe
>> 422>NAME STATUS WARN ASSERTS FAILED PERCNT
>> USER SYS REAL
>> 422>accumulate.exe 0 0 0.330
>> 422>example output differs from the expected
>> 422>example failed! Exit code 0
>> 422>
Re: failed examples reported as passing with MSVC or vice versa
Posted by Andrew Black <ab...@roguewave.com>.
Greetings Martin.
My first impression is that the problem with the windows infrastructure
rather than the exec utility. It should be noted that the build in
question (likely) is a 64 bit build, necessitating use of one of the
shared 64 bit build machines. I suspect the significant portion of the
log is the few lines following those quoted. Extracted from a candidate
build, the lines in question are
422>d:\bman5\builds\33262615\source-buildspace\etc\config\windows\runall.wsf(59,
6) Microsoft JScript runtime error: Permission denied
422>Project : error PRJ0019: A tool returned an error code from
"Performing Custom Build Step"
422>Build log was saved at
"file://D:\bman5\builds\33262615\source-buildspace\build\msvc-8.0-x64\15s\examples\BuildLog.htm"
422>.stdlib_examples - 1 error(s), 0 warning(s)
My reading of this message is that the runall.wsf script failed to
complete a file system command. Unfortunately, I am having difficulties
determining where the failure occurred, as the line number is
meaningless within the context of a text editor. I tend to suspect that
this translates to line 118 of the runall.wsf script. A potential
contributing cause is the permissions of the build directory used by the
nightly testing system.
--Andrew Black
Martin Sebor wrote:
> While going through our MSVC 8.0 jumbled up log I came across
> what looks like incorrect behavior of either the exec utility
> or the Windows infrastructure. What seems to be happening is
> that while exec reports some examples as passing the rest of
> the infrastructure actually says they failed. Below are lines
> from the log relevant to the accumulate.exe example.
>
> Andrew and/or Farid, could you guys look into it to see which
> part of the infrastructure is broken?
>
> Thanks
> Martin
>
> 422>Performing Custom Build Step
> 422>running accumulate.exe
> 422>NAME STATUS WARN ASSERTS FAILED PERCNT USER
> SYS REAL
> 422>accumulate.exe 0 0 0.330
> 422>example output differs from the expected
> 422>example failed! Exit code 0
> 422>