Re: [cxx-abi-dev] IEEE 754r half-precision fp mangling?
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [cxx-abi-dev] IEEE 754r half-precision fp mangling?
- To: scott douglass <scott.douglass@xxxxxxx>
- Subject: Re: [cxx-abi-dev] IEEE 754r half-precision fp mangling?
- From: Mark Mitchell <mark@xxxxxxxxxxxxxxxx>
- Date: Mon, 09 Jun 2008 14:30:59 -0700
scott douglass wrote:
I would suggest you just use "D". I.e. save "E" for End as in most
other parts of the BNF.
Thanks. I'm happy to use 'D<letter>' if there's already a precedent.
I don't see any precedent there -- but I think we could do that.
If 'Dh' is available, I change my proposal to:
<type> ::= Dh # half-precision floating point (16bit)
Here is another possible trick that we could use for both DFP and
half-precision float... Since we have:
::= u <source-name> # vendor extended type
and since a source name always starts with a decimal integer, we already
have a piece of the encoding space reserved, given by the regular
expression:
u[a-zA-Z].+
So, we could use "uh" for half-precision float. That's the same number
of characters and avoids using any new encoding space?
--
Mark Mitchell
CodeSourcery
mark@xxxxxxxxxxxxxxxx
(650) 331-3385 x713