Poll: Primary D version

[Jonathan M Davis]

Andrei Alexandrescu wrote:
> 
> You both have a point. Clearly not a lot of individual applications
> really need more than 4GB (though unfortunately, many are pushing up for
> the wrong reasons), but then a whole category of them would greatly
> benefit of expanded RAM availability.
> 
> Andrei

I've written at least one application (for my thesis) which ended up using 
all of my 4GB RAM and 6GB swap. Of course, that was at least partially 
because I was writing it in Haskell and hadn't taken its laziness into 
proper account. It was reading in the hundreds of files before it actually 
calculated anything since it didn't write anything to disk until it was done 
processing (which it naturally never did since it ran out of memory). Fixing 
it to write to disk after processing each file (thereby forcing it to 
actually process each file before reading in the next one) made it only take 
3+ GB of RAM. But I was doing a lot of string processing, and it wasn't at 
all a typical app. Haskell was a poor match for the problem as it turns out, 
but given D's current lack of 64-bit support, it would have been too - 
though for very different reasons.

Still, you work with what you've got. We'll get 64-bit support eventually. 
At least I can say that I wrote a program that used up all of my memory and 
swap doing something useful (or trying anyway). I don't think that many 
people can say that - especially when it was around 10GB total. That project 
definitely lead me to upgrade my RAM. But anywho, D is great. And for the 
most part, 64-bit isn't necessary. But it will be nice when we do get it.

- Jonathan M Davis