Re: [Boost-testing] Boost.Locale failure killing nightly tests

Subject: Re: [Boost-testing] Boost.Locale failure killing nightly tests
From: Belcourt, Kenneth (kbelco_at_[hidden])
Date: 2011-07-15 19:24:49

Hi Artyom,

On Jul 15, 2011, at 2:04 PM, Artyom Beilis wrote:

>> From: "Belcourt, Kenneth" <kbelco_at_[hidden]>
>> To: Running Boost regression tests <boost-testing_at_[hidden]>
>> Sent: Fri, July 15, 2011 8:08:36 PM
>> Subject: [Boost-testing] Boost.Locale failure killing nightly tests
>> 
>> I'm trying to standup a Pgi 11.7 tester but it's failing with this  error 
>> message.  Has something changed in Boost.Locale that would have  caused this?
>> 
>> [ excerpt from bjam.log in the test directory  ]
>> 
>> Performing configuration checks
>> 
>>    - has_icu  builds           : no
>>    - iconv  (libc)             : no
>>    - iconv  (separate)         : no
>>    - icu                       : no
>> error:  Boost.Locale needs either iconv or ICU library to be built.

I get the same problem with Mac OS X (Snow Leopard).

====================================================
notice: OSX version on this machine is 10.6.6
notice: will use '/opt/local/bin/g++-mp-4.4' for darwin, condition <toolset>darwin-4.4
notice: using strip for <toolset>darwin-4.4 at strip
notice: using archiver for <toolset>darwin-4.4 at libtool
notice: will use '/opt/local/bin/g++-mp-4.4' for darwin, condition <toolset>darwin-4.4_0x
notice: using strip for <toolset>darwin-4.4_0x at strip
notice: using archiver for <toolset>darwin-4.4_0x at libtool
notice: iostreams: using prebuilt zlib
notice: iostreams: using prebuilt bzip2
Performing configuration checks

    - has_icu builds           : no
    - iconv (libc)             : no
    - iconv (separate)         : no
    - icu                      : no
error: Boost.Locale needs either iconv or ICU library to be built.

> Actually it shows that the Boost.Locale library can't be build,
> because it requires either ICU or iconv, or win32 API to be available
> at least.
> 
> Two things:
> 
> 1. Why does it stop testing if one library can't be build?

No idea, perhaps because it's a failed configuration test rather than a failed normal test?

>   What can be done to fix it?

Sorry, no idea.

> 2. It is very strange that the system does not finds iconv - it is 
>   Linux and it is the a part of libc's API.

Yup, seems like your configuration check isn't working.

-- Noel