Re: [boost] Re: unit_test_development merge progress

From: William E. Kempf (williamkempf_at_[hidden])
Date: 2002-08-12 09:17:57

----- Original Message -----
From: "Beman Dawes" <bdawes_at_[hidden]>
To: <boost_at_[hidden]>; <boost_at_[hidden]>
Sent: Monday, August 12, 2002 8:52 AM
Subject: Re: [boost] Re: unit_test_development merge progress

> At 09:08 AM 8/12/2002, William Kempf wrote:
>
>  >Actually, there have been few maitenance things done to Boost.Threads,
> and
>  >those that were should pass testing on the compilers that passed
> before.
>  >Most of the recent work has actually be done to support the new unit
>  >testing
>  >framework, and I expect it to all work from the threads/build directory
> on
>  >the compilers where it worked before... but I'm sure I've broken the
>  >status/Jamfile stuff in the interim.  I hope to correct that today.
>
> By the time I read the above, I'd already guessed that was what you had
> done and added a line to status/Jamfile.

Actually, now that I get the opportunity to test on another machine I'm
finding some issues with recent changes as well.  I'll get the kinks worked
out today.

> That resulted in a link error
>
>    Unresolved external 'init_unit_test_suite(int, char * *)'

Did you check in the status/Jamfile?  I'll probably be able to fix that as
well.

> I've also emailed that to Gennadiy; I expect it will be simple for him to
> fix.
>
> One oddity that has affected several libraries, but not threads, is that
> Intel and VC++ 7.0 get link failures if test_main is defined like this:
>
>      int test_main(int,char**)
>
> I've just been changing those to
>
>      int test_main(int,char*[])
>
> which works fine.
>
> I think that what is happening is that when char*[] decays, a const is
> getting added.  I have no idea if that is standards conforming.

Interesting.  We'll have to check the standard, and if it's conforming we
should probably add the const ourselves, and if it's not, report the bug to
the vendors.

Bill Kempf