John,

> Or if you want to use a "real" language, use Felix, where "inline"
> isn't a hint.

Can't blame you for marketing your product, but does it run under the
NetBurner Eclipse IDE to produce code for a Coldfusion Freescale
processor?  :-)  And call C language library code from NB too?

And can you supply a C-to-Felix translator that's 100% bulletproof?
(Wouldn't surprise me if you could. :-)  And overcome the inertia
here among people resistant to new platforms and languages?

>> I vaguely recall (this was > 10 years ago) discussing multi-index
>> operations like this.  I know I actually WROTE code that batch-created
>> (inserted) a Judy1/L array rather faster than by simple iteration, and
>> it worked fine, but I don't know where it is today.

> Actually that would be quite useful.  I actually use Judy as a
> temporary data structure.  The GC algorithm is basically:  find all
> reachable objects, then delete everything not reachable.

Doug, can you send him the "valid" copy of that code, or should I try to
dig it up?  I don't even recall what it was called...  JudyLInsBatch(),
etc, maybe?

> All-reachable is a set of pointers (Judy) constructed by scanning.
> Everything-not-reachable is a set of pointers.  Also Judy.  This is
> constructed because the practice is to run finalisers on all these
> objects *first* before deallocation.

OK.

> Just an option for two words would be useful.
>
> Why?

Why stop at two words?  :-)

> Well, there is a serious bug/misconception in the Judy build.
> Actually, the Felix clone makes the same mistake.
>
> I'm running on a 32 bit machine.  I build only 32 bit Judy.  Stupid.
> I still need 64 bit integers.
>
> I'm running on a 64 bit machine.  I build 64 bit Judy.  Stupid.  I
> still need "int".  That's only 32 bits.  Heck, "long" is only 32 bits
> on Win64.

I see, having to bind the library to one size or the other is hobbling.

> What should be built is:  32 and 64 bit arrays.  On BOTH 32 and 64 bit
> machines.

I think you can, but not sure how they'd coexist -- namespace issues.

> In the C++ Standard they're called NTBS:  Null terminated byte
> strings.  So that's a technical, formal, and standardised name for "C
> strings" :-)

I'll try to remememember that.  :-)

> This is similar to the property naive Hash-table has.  You lose
> ordering in return for O(1) operations.

Yup.

Cheers,
Alan

------------------------------------------------------------------------------
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data applications. Try Boundary one-second 
resolution app monitoring today. Free.
http://p.sf.net/sfu/Boundary-dev2dev
_______________________________________________
Judy-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/judy-devel

Reply via email to