James Carlson <carlsonj at workingcode.com> writes:
>> To be honest, I myself did not yet read this TR, and that in particular is
>> due to the
>> fact that we did not have any requests to support it till this very moment.
>>
>> Eventually we will, but we need to prioritize.
>> And getting external interest for the feature helps putting proper
>> priorities.
>
> No doubt. And to amplify the point: none of the organizations here are
> operating as charities, so if there are paying customers asking through
> the appropriate channels (their local sales and support folks) to have
> work prioritized, that goes *much* further to getting work funded.
>
> Apart from that, there is another path. Anyone who wants to can put
> together the proposed new ABI, modify some open source compiler to
> comply, and run the architectural bits through the ARC. If the Sun
> Studio team can contribute, that's great, but if that team's priorities
> are elsewhere, I see nothing at all wrong with having the community
> members with an interest in this driving the effort rather than just
> "asking" for the work to be done.
Agreed: my mail didn't intend to demand the work being done by the
Studio and libc teams, but to determine if there was ongoing work in the
area. It would be foolish to start anything in this area without first
making sure that nobody else went there before.
That said, I doubt that I've got the spare time to define a decimal
float on SPARC ABI myself and drive that through SPARC International for
acceptance. After all, both my GCC and OpenSolaris work are spare-time
activities. What I might try is use the established x86/amd64 ABI and
try to get GCC to support decimal float on Solaris/x86 and see if I can
adapt existing code with appropriate licenses (i.e. not GPL) for the
necessary libc additions.
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University
