Jacob Carlborg doob at me.com
Sat Feb 11 04:39:16 PST 2012

On 2012-02-10 23:06, Jonathan M Davis wrote:
> On Friday, February 10, 2012 13:54:49 H. S. Teoh wrote:
>> Perhaps one solution is to simply have the compiler recognize all valid
>> D constructs, but to give an error when it sees a construct not
>> supported by the target platform. For example:
>> input.d(123): Error: dynamic arrays not supported by target
>> platform.
>> This way, we don't cause fragmentation of D, and at the same time users
>> will know which constructs should be avoided due to target platform
>> limitations.
> I think that the only switch even vaguely like that which has any chance of
> making it into the compiler is one which makes any constructs which use the GC
> illegal. And no one has been able to get Walter to even put _that_ in yet.
> For the most part, I'd argue that if you don't want to use all of D's
> features, then just don't use them all. The only major issue that I'm aware of
> is if you want to avoid the GC entirely (either for performance reasons or
> paranoia), you're probably going to have to forgoe using a lot of Phobos.
> - Jonathan M Davis

It's possible to remove the GC and you'll get linker errors when the GC 
is used. This is a hack but it works today.

/Jacob Carlborg

More information about the Digitalmars-d mailing list