Who still use a 32-bit computer these days anyways.

Daniel Ribeiro Maciel wrote:
> Who still use a 32-bit computer these days anyways.

Most people.

Daniel Ribeiro Maciel wrote:
> Who still use a 32-bit computer these days anyways.

I do.

On Sat, 01 Nov 2008 15:41:16 +0300, Daniel Ribeiro Maciel <daniel.maciel@gmail.com> wrote:

> Who still use a 32-bit computer these days anyways.

I bet most Windows users are using 32bit version of Windows, even if their CPUs are x64-capable.

Daniel Ribeiro Maciel wrote:
> Who still use a 32-bit computer these days anyways.

I use 32bit Windows ('cause it is bundled) and 64bit Linux.

BTW, gdc is 64 bit (x86-64), though it only supports D1.

Robert Fraser wrote:
> Daniel Ribeiro Maciel wrote:
>> Who still use a 32-bit computer these days anyways.
> 
> Most people.
Since the topic is here, I'd like to make a suggestion for when to make a 64-bit compiler.
I think the right time is when D 2.0 is stable and we begin to move away from D 1.0. Instead of jumping straight to D 3.0, we could spend a little time developing a 64-bit compiler for D. It would give us a stable and unstable branch like we have already. And developers could update their code for D 2.0 and 64-bit hardware at the same time.
Besides, I don't think we should start development on D 3.0 so soon. I think it's better if a language doesn't evolve too quickly. It gives time for developers to get settled in and write stable code for D, without the need to update it every couple of years.

== Quote from Xinok (xnknet@gmail.com)'s article
> Robert Fraser wrote:
> > Daniel Ribeiro Maciel wrote:
> >> Who still use a 32-bit computer these days anyways.
> >
> > Most people.
> Since the topic is here, I'd like to make a suggestion for when to make
> a 64-bit compiler.
> I think the right time is when D 2.0 is stable and we begin to move away
> from D 1.0. Instead of jumping straight to D 3.0, we could spend a
> little time developing a 64-bit compiler for D. It would give us a
> stable and unstable branch like we have already. And developers could
> update their code for D 2.0 and 64-bit hardware at the same time.
> Besides, I don't think we should start development on D 3.0 so soon. I
> think it's better if a language doesn't evolve too quickly. It gives
> time for developers to get settled in and write stable code for D,
> without the need to update it every couple of years.

Given that LDC is going to be released any day now, and should support 64-bit, any estimates of how long it will take for LDC to support D2?  I work with huge datasets and occasionally run up against the 2 gig address space barrier of 32 bit on a computer with 4+ gigs of RAM.  Therefore, I for one would really find such a thing useful.

On Sat, 01 Nov 2008 16:09:39 +0300, Xinok <xnknet@gmail.com> wrote:

> Robert Fraser wrote:
>> Daniel Ribeiro Maciel wrote:
>>> Who still use a 32-bit computer these days anyways.
>>  Most people.
> Since the topic is here, I'd like to make a suggestion for when to make a 64-bit compiler.
> I think the right time is when D 2.0 is stable and we begin to move away from D 1.0. Instead of jumping straight to D 3.0, we could spend a little time developing a 64-bit compiler for D. It would give us a stable and unstable branch like we have already. And developers could update their code for D 2.0 and 64-bit hardware at the same time.
> Besides, I don't think we should start development on D 3.0 so soon. I think it's better if a language doesn't evolve too quickly. It gives time for developers to get settled in and write stable code for D, without the need to update it every couple of years.

Walter is concentrated on the language (specs and a reference implementation).
There is, however, a compiler at works that will have 64bit codegen one day:

    http://www.dsource.org/projects/ldc

Daniel Ribeiro Maciel wrote:
> Who still use a 32-bit computer these days anyways.

Donald Knuth does.  He's complained about not being able to leverage the additional registers in a 64-bit CPU without having to use 64-bit pointers though.  See:

http://www-cs-staff.stanford.edu/~uno/news.html

"A Flame About 64-bit Pointers"


Sean

> Given that LDC is going to be released any day now, and should support 64-bit

I have to correct that. While we hope to release LDC for x86-32 Linux soon after LLVM 2.4 is released, x86-64 (and, in fact, any other platform) still needs work.

K. Wilson has contributed some big fixes for x86-64 Linux recently (he made inline asm and exception handling work! yay for him!) and reports that things basically work, but there's still lots of testing - and most likely bugfixing - to be done before it's ready.

I've already said it about LDC x86-32 Windows, but it's the same here: we could really use a few more dedicated x86-64 people!


> any estimates of how long it will take for LDC to support D2?

Neither Tomas nor me are particularly interested in chasing D2 right now. Providing a cross-platform D1 compiler and fixing bugs there seems more useful at the moment.