Does the regex lib go on to build OK?
Those compiler errors are from a configuration check that looks to see if
ICU is present or not - you shouldn't actually see those messages at all
(they should be suppressed and written to file rather like autoconf does).
Is there a bjam command line that reliably reproduces this?
Thanks, John.
_______________________________________________
Boost-users mailing list
Boost...@lists.boost.org
http://lists.boost.org/mailman/listinfo.cgi/boost-users
>> I have written a bash build script do this build. It has worked fine for
>> previous version of boost.
>> i.e. 1.37.0, 1.40.0, 1.43.0
>> But when I tried to build 1.44.0 I get the following errors.
>
> Does the regex lib go on to build OK?
>
> Those compiler errors are from a configuration check that looks to see if
> ICU is present or not - you shouldn't actually see those messages at all
> (they should be suppressed and written to file rather like autoconf does).
>
> Is there a bjam command line that reliably reproduces this?
I think it's the same 'config checks on fresh build go to stdout' bug that
I've mentioned before. I hope to fix it soonish as part of Boost.Locale
configuration work.
- Volodya
Does the regex lib go on to build OK?I have written a bash build script do this build. It has worked fine for
previous version of boost.
i.e. 1.37.0, 1.40.0, 1.43.0
But when I tried to build 1.44.0 I get the following errors.
Those compiler errors are from a configuration check that looks to see if ICU is present or not - you shouldn't actually see those messages at all (they should be suppressed and written to file rather like autoconf does).
Is there a bjam command line that reliably reproduces this?
Thanks, John.
John Maddock wrote:I think it's the same 'config checks on fresh build go to stdout' bug that
>> I have written a bash build script do this build. It has worked fine for
>> previous version of boost.
>> i.e. 1.37.0, 1.40.0, 1.43.0
>> But when I tried to build 1.44.0 I get the following errors.
>
> Does the regex lib go on to build OK?
>
> Those compiler errors are from a configuration check that looks to see if
> ICU is present or not - you shouldn't actually see those messages at all
> (they should be suppressed and written to file rather like autoconf does).
>
> Is there a bjam command line that reliably reproduces this?
I've mentioned before. I hope to fix it soonish as part of Boost.Locale
configuration work.
- Volodya
John Maddock wrote:I think it's the same 'config checks on fresh build go to stdout' bug that
>> I have written a bash build script do this build. It has worked fine for
>> previous version of boost.
>> i.e. 1.37.0, 1.40.0, 1.43.0
>> But when I tried to build 1.44.0 I get the following errors.
>
> Does the regex lib go on to build OK?
>
> Those compiler errors are from a configuration check that looks to see if
> ICU is present or not - you shouldn't actually see those messages at all
> (they should be suppressed and written to file rather like autoconf does).
>
> Is there a bjam command line that reliably reproduces this?
I've mentioned before. I hope to fix it soonish as part of Boost.Locale
configuration work.
- Volodya
The workaround is "mkdir <boost-root>/bin.v2"
HTH,
Does the regex lib go on to build OK?I have written a bash build script do this build. It has worked fine for
previous version of boost.
i.e. 1.37.0, 1.40.0, 1.43.0
But when I tried to build 1.44.0 I get the following errors.
Those compiler errors are from a configuration check that looks to see if ICU is present or not - you shouldn't actually see those messages at all (they should be suppressed and written to file rather like autoconf does).
Is there a bjam command line that reliably reproduces this?
Thanks, John.
> On Sat, Sep 25, 2010 at 4:45 AM, John Maddock <boost...@virgin.net>wrote:
>
>> I have written a bash build script do this build. It has worked fine for
>>> previous version of boost.
>>> i.e. 1.37.0, 1.40.0, 1.43.0
>>> But when I tried to build 1.44.0 I get the following errors.
>>>
>>
>> Does the regex lib go on to build OK?
>>
>> Those compiler errors are from a configuration check that looks to see if
>> ICU is present or not - you shouldn't actually see those messages at all
>> (they should be suppressed and written to file rather like autoconf does).
>>
>> Is there a bjam command line that reliably reproduces this?
>>
>> Thanks, John.
>> ost-users <http://lists.boost.org/mailman/listinfo.cgi/boost-users>
I think you misread this output. The error message is from the configuration check --
exactly the check that causes "has_icu builds" to output "no". Besides slight
inconvenience, this should not cause any problems, and regex should be built,
albeit without ICU support.
Do you actually observe that regex is not built? If so, there must be some further
error messages. Would you please put a complete build log somewhere?
- Volodya
Are you sure? What are the error messages?
> The compiler error:
> libs/regex/build/has_icu_test.cpp:12:30: error: unicode/uversion.h: No
> such
> file or directory
> is from trying to compile has_icu_test.cpp with my cross-compiler tools
> which do not have the icu libraries and header files installed.
> So it can not find unicode/uversion.h.
>
> Why is it trying to compile has_icu_test.cpp when has_icu builds = no, as
> shown below in my output?
Compiling has_icu_test.cpp is the test that's carried out when performing
the configuration checks - those errors should be directed to file and not
console - that's the bug that's being investigated.
Are you sure? What are the error messages?No, the regex lib does not go on to build.
Compiling has_icu_test.cpp is the test that's carried out when performing the configuration checks - those errors should be directed to file and not console - that's the bug that's being investigated.
The compiler error:
libs/regex/build/has_icu_test.cpp:12:30: error: unicode/uversion.h: No such
file or directory
is from trying to compile has_icu_test.cpp with my cross-compiler tools
which do not have the icu libraries and header files installed.
So it can not find unicode/uversion.h.
Why is it trying to compile has_icu_test.cpp when has_icu builds = no, as
shown below in my output?
John.
On 9/28/2010 7:59 AM, Robert McCullough wrote:
> I found my problem.
> I was building boost with the bjam -q option.
> When I build using the bjam -q option the build stops on the above error and
> does not build the libraries.
> I removed the -q and it ignored the above error and everything built fine.
For the record, this has been fixed in the trunk.
In Christ,
Steven Watanabe