[M3devel] LONGINT on NT386

[Tony Hosking]

On 12 Jan 2010, at 22:27, Jay K wrote:

> It's not easy to fix.
> I386_MINGWIN is the easiest way actually..
> I'll look at it again soon..
> Maybe we should have an option where the front end decomposes things?

It is the job of the backend to map to machine types.  Don't go complicating the front-end unnecessarily: compiler design 101.

> Or makes function calls?
> That is probably a pleasantly nice option.
> 
> 
> - Jay
> 
> 
> ________________________________
>> From: rcolebur at SCIRES.COM
>> To: m3devel at elegosoft.com
>> Date: Tue, 12 Jan 2010 21:57:02 -0500
>> Subject: [M3devel] LONGINT on NT386
>> 
>> 
>> Ok, so now that we have this LONGINT type, what else needs
>> to take place for it to be useful?
>> 
>> I ask, because currently on Windows, LONGINT has same range
>> as INTEGER.
>> 
>> 
>> Is there anything I can do to help?
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> BYTESIZE(INTEGER) =4
>> 
>> BYTESIZE(CARDINAL)=4
>> 
>> BYTESIZE(LONGINT) =4
>> 
>> 
>> 
>> FIRST(INTEGER) =-2147483648
>> 
>> LAST (INTEGER) = 2147483647
>> 
>> 
>> 
>> FIRST(CARDINAL)=0
>> 
>> LAST (CARDINAL)=2147483647
>> 
>> 
>> 
>> FIRST(LONGINT) =-2147483648
>> 
>> LAST (LONGINT) = 2147483647
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> Regards,
>> 
>> 
>> 
>> Randy Coleburn
>> 
>>