DMD and 64-bit

[Daniel Gibson]

Am 20.02.2011 03:49, schrieb Jonathan M Davis:
> On Saturday 19 February 2011 18:27:17 Daniel Gibson wrote:
>> Am 20.02.2011 03:22, schrieb Jonathan M Davis:
>>> On Saturday 19 February 2011 16:21:05 Nick Sabalausky wrote:
>>>> "Jonathan M Davis"<jmdavisProg at gmx.com>   wrote in message
>>>> news:mailman.1787.1298108224.4748.digitalmars-d at puremagic.com...
>>>>
>>>>> On Saturday 19 February 2011 01:12:25 Russel Winder wrote:
>>>>>> Am I correct in assuming that DMD generates 32-bit by default and that
>>>>>> for 64-bit you have to give the -m64 option?
>>>>>>
>>>>>> Is the eventual plan to use the natural word length of the platform as
>>>>>> the default, i.e. 32-bit on 32-bit and 64-bit on 64-bit, with the
>>>>>> option to force something different using the -m option à la GCC?
>>>>>
>>>>> 32 bit is the default for now. I don't know what Walter intends to do
>>>>> in the
>>>>> long run. If it were like gcc, then the default platform would be the
>>>>> platform
>>>>> which the compiler is built for (which is exactly what it's doing right
>>>>> now),
>>>>> but as far as I know, Walter has no plans to port dmd to x86_64.
>>>>>
>>>>> So, for now 32 bit is the default. It may or may not change later to be
>>>>> the
>>>>> platform that you're building on. Ideally though, it would be like gcc
>>>>> and we'd
>>>>> actually get a 64 bit version of dmd at some point.
>>>>
>>>> I don't mean this as "anti-64bit" trolling, just a genuine question, but
>>>> what would be the point of a 64-bit build of DMD? (Note: I'm talking
>>>> about host, not target). Just to compile projects that have (on the
>>>> order of) gigabytes of source?
>>>
>>> 1. Pure 64-bit systems don't run 32-bit programs unless they're
>>> statically linked. You can't currently use dmd on a pure 64-bit system.
>>> You either have to have a multilib system (which I'd argue you really
>>> should be doing anyway, but it can be annoying) or you have to have a
>>> chrooted 32-bit environment to run it in (yuck). And even if you _can_
>>> have a multilib system, it's still nice to have as many programs
>>> natively 64-bit as possible - especially since it reduces the number of
>>> stray 32-bit libraries that you have to have installed.
>>>
>>> 2. dmd uses a lot of memory. It's trivial to write a program which causes
>>> dmd to run out of memory - especially when using a lot of templates or
>>> CTFE. Now, a good chunk of that is caused by dmd bugs and not an
>>> intrinsic need for more memory, but it could still be very useful to
>>> have dmd be able to go beyond the 32-bit memory barrier.
>>
>> This would however promote the development of programs that can be
>> compiled on (for?) 64bit only. Not sure if this is desirable, especially
>> as long as there's only 64bit support on Linux but not Windows/OSX (What
>> about FreeBSD? Is it supported yet? Is it much different from Linux in
>> this regard?)
>
> Is there much difference between making it so that you can have programs which
> can only be compiled on a 64-bit platform and programs which can only be run on
> a 64-bit platform? One of the big reasons that people want 64-bit support is
> precisely so that they can run programs which _can't_ run within the memory
> limits that 32-bit platforms have. The only difference here would be that then
> you'd have programs which couldn't be compiled on a 32-bit system. And
> particularly once CTFE is to the point that it can do all of SafeD (as is the
> eventual goal), then _of course_ you're going to get into situations where a 64-
> bit program won't compile in 32-bit land.
>
> Now, given the fact that dmd is so poor with memory, having a 64-bit version at
> this point would make it too easy to write a program which is _intended_ to work
> on both 32- bit and 64-bit systems only compilable on 64-bit systems,

This is what I meant :-)

> so that
> would be bad, but once dmd's memory problems have been solved, that wouldn't be
> an issue, and I would hope that fixing dmd's memory problems would be a higher
> priority than getting a dmd 64-bit executable.

That would definitely make sense.

>
> As for the state of FreeBSD, I don't know. FreeBSD isn't exactly well-tested in
> general, so even if it's _supposed_ to work, it may not. And there isn't
> currently an actual FreeBSD release of dmd, so I have no idea what it's state
> is. It may or may not be supposed to work with 64-bit right now. Walter does
> have some interest in FreeBSD, but when very few people around here use it and
> we have no autotester for FreeBSD, it makes it kind of hard to test.

Every once in a while somebody who compiles dmd on FreeBSD himself turns up.
There's even a Port: 
http://www.freebsd.org/cgi/ports.cgi?query=dmd&stype=all

>
> - Jonathan M Davis

Cheers,
- Daniel