Re: How to implement dynamic typing?

On 14 Apr 2010 17:55:11 +0100 (BST), Tony Finch <dot@dotat.at> wrote:


>George Neuner <gneuner2@comcast.net> wrote:
>>
>LuaJIT takes an interesting approach. Its only numeric type is the
>double precision floating point number. It uses different kinds of NAN
>as tags for the other types, with the pointer embedded in the
>mantissa.
>
>http://article.gmane.org/gmane.comp.lang.lua.general/44823
>
>Regarding typed word-sized pointers, see for instance
>http://www.haskell.org/~simonmar/papers/ptr-tagging.pdf


That is interesting. It does, however, prevent loading tagged values
directly as doubles if the CPU might throw an exception on a NAN. Also
the IEEE storage format for doubles leaves the tag in the middle of
the first word (the sign bit is first - NANs are encoded in the
exponent). That makes tag checking/extraction cumbersome on a 32-bit
machine.


George

Return to the comp.compilers page.
Search the comp.compilers archives again.