Newbie Q:

During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.

Any ideas?

Brian Hay wrote:
> Newbie Q:
> 
> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
> 
> Any ideas?

I think there's a limit of 16K for the size of static arrays.
Do you have any really big static arrays?

Anyway, the compiler should generate a line number for that.  Not providing one is a bug.

--bb

Bill Baxter wrote:
> Brian Hay wrote:
>> Newbie Q:
>>
>> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>>
>> Any ideas?
> 
> I think there's a limit of 16K for the size of static arrays.
> Do you have any really big static arrays?

No. Nothing like that.

> Anyway, the compiler should generate a line number for that.  Not providing one is a bug.

That's what I figured. It's kind of hard to track down the issue without such details.

"Brian Hay" <bhay@construct3d.com> wrote in message news:frviab$126$1@digitalmars.com...
> Newbie Q:
>
> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>
> Any ideas?

Hee hee hee.

static int[] x = [-1: 1];

There's some kind of funky logic in the compiler which for some reason does not detect negative initialization indices, but rather does something like

currentLength = index.toInt();
currentLength++;

if(currentLength == 0)
    error("array dimension overflow");

It's to detect index overflow, but it's easily fooled by using a negative index, which it should really detect anyway.

Are you using, maybe, 0xFFFF_FFFF anywhere as an array initializer index? Also, are you getting an "array initializers as expressions are not allowed" error at the same time by any chance?

Jarrett Billingsley wrote:
> "Brian Hay" <bhay@construct3d.com> wrote in message news:frviab$126$1@digitalmars.com...
>> Newbie Q:
>>
>> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>>
>> Any ideas?
> 
> Hee hee hee.
> 
> static int[] x = [-1: 1];

Why do you hate the compiler? Meanie.

/me sniffles.

Jarrett Billingsley wrote:
> "Brian Hay" <bhay@construct3d.com> wrote in message news:frviab$126$1@digitalmars.com...
>> Newbie Q:
>>
>> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>>
>> Any ideas?
> 
> Hee hee hee.
> 
> static int[] x = [-1: 1];
> 
> There's some kind of funky logic in the compiler which for some reason does not detect negative initialization indices

Thanks Jarrett, that was it.

== Quote from Brian Hay (bhay@construct3d.com)'s article
> Bill Baxter wrote:
> > Brian Hay wrote:
> >> Newbie Q:
> >>

...
> > Anyway, the compiler should generate a line number for that.  Not providing one is a bug.
>
> That's what I figured. It's kind of hard to track down the issue without such details.

I agree.

I filed a bug report so that perhaps a future version of DMD would include a line number when it runs into this error: http://d.puremagic.com/issues/show_bug.cgi?id=1936

Bill Baxter wrote:
> Brian Hay wrote:
>> Newbie Q:
>>
>> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>>
>> Any ideas?
> 
> I think there's a limit of 16K for the size of static arrays.
> Do you have any really big static arrays?
> 
> Anyway, the compiler should generate a line number for that.  Not providing one is a bug.
> 
> --bb
16k Seems shockingly low. Is there documentation on this. Even old Dos compilers for C support close to 64k indexes.

"Spacen Jasset" <spacen@yahoo.co.uk> wrote in message news:ft35ik$q7g$1@digitalmars.com...
> Bill Baxter wrote:
>> Brian Hay wrote:
>>> Newbie Q:
>>>
>>> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>>>
>>> Any ideas?
>>
>> I think there's a limit of 16K for the size of static arrays. Do you have any really big static arrays?
>>
>> Anyway, the compiler should generate a line number for that.  Not providing one is a bug.
>>
>> --bb
> 16k Seems shockingly low. Is there documentation on this. Even old Dos compilers for C support close to 64k indexes.

"Jarrett Billingsley" <kb3ctd2@yahoo.com> wrote in message news:ft3757$11cb$1@digitalmars.com...
> "Spacen Jasset" <spacen@yahoo.co.uk> wrote in message news:ft35ik$q7g$1@digitalmars.com...
>> Bill Baxter wrote:
>>> Brian Hay wrote:
>>>> Newbie Q:
>>>>
>>>> During compilation I'm getting the error: "array dimension overflow" but it doesn't provide any information as to where (module and line number) the error might be, so I'm at a loss. Other errors seem to provide such information.
>>>>
>>>> Any ideas?
>>>
>>> I think there's a limit of 16K for the size of static arrays. Do you have any really big static arrays?
>>>
>>> Anyway, the compiler should generate a line number for that.  Not providing one is a bug.
>>>
>>> --bb
>> 16k Seems shockingly low. Is there documentation on this. Even old Dos compilers for C support close to 64k indexes.

Son of a.

Bill made a typo, the max is 16MB.  While Walter says that this is because "any larger array should really be allocated at runtime and not in the static data segment", I think it's really because OPTLINK crashes if you have something larger than 16MB in the static data segment.

9_9