On 09/04/12 05:37, Stefan Teleman wrote:
> On Mon, Sep 3, 2012 at 11:57 PM, Stefan Teleman
> <stefan.teleman@gmail.com> wrote:
>
>> http://old.nabble.com/22.locale.numpunct.mt-run-hangs-td20133013.html
>
> Also see this thread:
>
> http://www.mail-archive.com/issues@stdcxx.apache.org/msg01182.html
>
> and this:
>
> https://issues.apache.org/jira/browse/STDCXX-839

Hi Stefan,

IMHO Martin's report with the crashing test as well as your with the apparent hanging of the Intel build deserve more attention than other bug reports. A test timing out, regardless of the CPU activity, is more indicative of an inefficiency. All my tests were timing out

Again, my concern is that a bug in the reference counting of std::string (or locale facets management) should be reproducible everywhere. If not, then only a subset of platforms are at fault -- so what is it? Is the atomic ops code defective there? We should be able to pinpoint the defect accurately.

By no means I am dismissing it. It is at the very least an issue of efficiency. I will try an Intel C++ build on x86_64 at some point today. What build type was that? I also notice that you have 4.2.1 in your path? Are you building out of 4.2.1. tag? I built off 4.2.x branch which also has support for custom timeouts (--soft-timeout) in rw_test.

Thanks!

Liviu