Re: [Felix-language] [felix] JudySL problem

[john s. <sk...@us...>]

On 01/11/2011, at 1:13 AM, Doug Baskins wrote:

> John:
> 
> My latest version of Judy checks to see if it is being compiled in a
> 64 bit environment and sets the JU_64BIT macro.  Sorry I did not
> do that in the released version.

I'm not sure that's a good idea. How are you going to get this right?
Note that just because the host machine is 64 bit doesn't mean the target
is also 64 bit: for example compiling for iPhone (ARM processor!) on
modern Mac (Intel, 64 bit).

I wish you'd get rid of the table gen stuff. There are only two sizes in reality,
32 and 64 bit. Why not just commit both to the repository, already generated?

The problem is the generator must NOT use the environment size as that
of the generated tables. The generator runs on the host machine, the tables
may be used on Judy running on a completely different machine.

The same applies to endian-ness.


>  I wish the compilers had a portable
> way of testing for a 32/64 bit compile in the code. 

What? Expect a Standards committee to do something useful?

> Good catch John Tromp.  I would of never found that one.

I suspected it, because I knew Erick changed the Felix build process for Judy
from my original one which used modified files, to one which used the
authentic originals from sourceforge -- thereby allowing any upgrades 
to be imported.

Also the corruption symptoms suggested it, since I know Judy is reliable
if built properly.

Thanks everyone for your attention! It remains to get Erick to fix the build
script (since I can't figure out how to do it).

--
john skaller
sk...@us...