Re: [Boost-testing] VC8 results from very recent compiler

From: Carl Daniel (cpdaniel_at_[hidden])
Date: 2005-10-09 23:36:58

Jeff Garland wrote:
> On Sat, 8 Oct 2005 17:37:07 -0700, Carl Daniel wrote
>> Results from testing VC8 build 14.00.50727.42 were just uploaded.
>> While the official release to manufacturing hasn't occurred yet,
>> everyone I've asked inside Microsoft says that they expect this to
>> be the final build before RTM.  This is about 2 weeks newer than the
>> "Release Candidate" that was handed out at the PDC.
>
>
> There's still a runtime failure in the date_time testlocal_time test
> -- so I guess it may be time to spend more time looking at a work
> around.  Any chance I can get you to put this into the debugger and
> send me a stack trace.

I assume there's a single .exe that I could run under teh debugger to repeat 
this test?  If so, just let me know which one and I'll see what I can do. 
Assume I no nothing about how the test framework works - and you won't be 
far off!

> Also, there are 2 other tests that passed,
> but are marked for failure because in previous beta's they've been
> failing -- we'll need to watch those because in earlier releases they
> wouldn't fail on every run...
>
> The other question, is what we should do with the deprecation
> warnings:
>
> E:\cpdaniel\Dev\boost_regression\boost\boost/date_time/c_time.hpp(68)
> : warning C4996: 'localtime' was declared deprecated
>        D:\Program Files\Microsoft Visual Studio
> 8\VC\INCLUDE\time.inl(114) : see declaration of 'localtime'
>        Message: 'This function or variable may be unsafe. Consider
> using localtime_s instead. To disable deprecation, use
> _CRT_SECURE_NO_DEPRECATE. See online help for details.'
> ...
>
> Since these are standard interfaces that have been 'deprecated' I'm
> tempted to say that since Boost believes in using interfaces in the
> standard, we should set _CRT_SECURE_NO_DEPRECATE in the config.
> Downside is that this is a big hammer...so maybe it should be done in
> the affected places.  Bottom line is I don't believe these warnings
> should appear just b/c you decide to use date-time in your
> program...Thoughts?

I'm on the fence, but I'd lean towards defining _CRT_SECURE_NO_DEPRECATE by 
default.  While their intentions were in the right place, I personally don't 
think the VC++ libraries team is going to make many friends with these 
deprecation warnings for standard-mandated functions.

-cd