Re: AIX testsuite failure due to missing template instantiation

[Phil Edwards <phil at jaj dot com>]

On Mon, May 12, 2003 at 02:45:57PM -0400, David Edelsohn wrote:
> >>>>> Phil Edwards writes:
> Phil> On Mon, May 12, 2003 at 11:14:01AM -0500, Benjamin Kosnik wrote:
> >> >FAIL: 27_io/basic_filebuf/underflow/10096.cc (test for excess errors)
> >> >Excess errors:
> >> >ld: 0711-317 ERROR: Undefined symbol: std::basic_filebuf<char, MyTraits>::_S_pback_size
> >> 
> >> Should be fixed now, thanks.
> 
> Phil> Is there a way for an ix86-linux user/maintainer to simulate the AIX
> Phil> no-weak no-common no-whatever situation, and find these things ahead of time?
> 
> 	-fno-implicit-templates

Oh, right, duh.


> We have decided that libstdc++ should be tested as it is run by the target
> -- using implicit template instantiation if the target supports that
> correctly.  One needs to explicitly use the option to check if a new or
> changed test will fail on targets without that functionality.

Yeah, it's coming back to me now.

Since you tend to find these errors before the rest of us, would you
consider making patches of the form


    * testsuite/........cc:  Add weak bits.

    #if !__GXX_WEAK__
    // Explicitly instantiate for systems with no COMDAT or weak support.
    template .....;
    #endif

?  I would consider that applicable under the obvious rule.  Opinions?

I made a couple attempts at running the testsuite in a streamlined "just
link, don't run" mode in order to try and /quickly/ test such instantiations,
but can't convince dejagnu to /not/ run them.


Phil

-- 
If ye love wealth greater than liberty, the tranquility of servitude greater
than the animating contest for freedom, go home and leave us in peace.  We seek
not your counsel, nor your arms.  Crouch down and lick the hand that feeds you;
and may posterity forget that ye were our countrymen.            - Samuel Adams