atom feed19 messages in net.java.dev.rococoa.usersRe: Stumped by 64-bit/1.6 Java error
FromSent OnAttachments
Matt ReynoldsSep 17, 2008 12:49 pm 
Duncan McGregorSep 18, 2008 6:10 am 
Paul LoySep 18, 2008 7:08 am 
Paul LoySep 18, 2008 7:08 am 
Matt ReynoldsSep 18, 2008 7:55 am 
Matt ReynoldsSep 18, 2008 1:50 pm 
Matt ReynoldsSep 18, 2008 4:59 pm 
Paul LoySep 19, 2008 3:40 pm 
Duncan McGregorSep 19, 2008 3:43 pm 
Matt ReynoldsSep 19, 2008 3:44 pm 
Paul LoySep 19, 2008 5:02 pm 
Duncan McGregorSep 20, 2008 2:33 am 
Paul LoySep 21, 2008 9:52 am 
Paul LoySep 21, 2008 9:53 am 
Paul LoySep 21, 2008 10:29 am 
Paul LoySep 21, 2008 11:47 am 
Paul LoySep 21, 2008 11:48 am 
Harald KuhrSep 22, 2008 1:58 am 
Paul LoySep 22, 2008 3:17 am 
Subject:Re: Stumped by 64-bit/1.6 Java error
From:Paul Loy (pa@keteracel.com)
Date:Sep 21, 2008 10:29:34 am
List:net.java.dev.rococoa.users

Ok, so, I can fiddle it at the moment, by changing all the Integer stuff from RococoaTypeMapper to Long stuff (i.e. long, longValue, etc). When I try to substitute NativeLong (a better solution) JNA tells me where to go. Now, as NativeLong will work for both 32 and 64, we can be doing this in parallel Mr. McGregor. No need for new hardware!

Paul.

On Sun, Sep 21, 2008 at 5:53 PM, Paul Loy <pa@keteracel.com> wrote:

yeah, ok, ID is different from id. My mistake. id is 8 bytes. Just like void *.

I think I've got a hangle on other problems. RococoaTypeMapper seems to be dealing in integers when it might want to be dealing with longs. Will look into that now.

Paul.

On Sat, Sep 20, 2008 at 10:34 AM, Duncan McGregor <dun@oneeyedmen.com>wrote:

I am a little surprised by sizeof(ID) being 4. What is sizeof(void*)?

My reading of, say,

http://developer.apple.com/documentation/MacOSX/Conceptual/OSX_Technology_Overview/PortingTips/chapter_9_section_2.html

is that it should be 8, but then that would mean that usual Cocoa casts like

id anID = ...

NSString* aString = (NSString*) anID;

would be suss.

Please don't make me buy a MacBook in the next week!

And thanks for the 2 of you making progress on this.

D

On 20 Sep 2008, at 01:03, Paul Loy wrote:

It may be too late, but I'm now a little confused. In 64-bits we have:

sizeof(ID) = 4 sizeof(SEL) = 8 sizeof(Class) = 8 sizeof(CFTypeRef) = 8

on 32-bit we have:

sizeof(ID) = 4 sizeof(SEL) = 4 sizeof(Class) = 4 sizeof(CFTypeRef) = 4

So I guess ID doesn't want to be a NativeLong? And then classes should not be IDs?

I need to sleep on this and then look at it again. (oh the CFTypeRef is the return type of CFRetain, we're currently invoking void on that one, when it does in fact have a return value of whatever you sent to CFRetain)

Much to ponder,

good night all

On Fri, Sep 19, 2008 at 11:45 PM, Matt Reynolds <ma@hom3.us> wrote:

Woo hoo! On to the next bug :)

--

--------------------------------------------- Paul Loy pa@keteracel.com http://www.keteracel.com/paul

--

--------------------------------------------- Paul Loy pa@keteracel.com http://www.keteracel.com/paul

--

--------------------------------------------- Paul Loy pa@keteracel.com http://www.keteracel.com/paul