Acorn Arcade forums: Programming: OS5 Window problem
| |
OS5 Window problem |
| |
pfj (14:52 18/1/2004)
adrianl (05:04 20/1/2004)
adrianl (04:36 25/1/2004)
pfj (21:20 31/1/2004)
|
| |
|
Paul F. Johnson |
Message #49785, posted by pfj at 14:52, 18/1/2004 |
Member
Posts: 54
|
I have an app which is compiled with a 32 bit clean version of RISC_OSLib.
It works fine on all machines except the Iyonix where it will install on the iconbar, but then all windows which open result in an illegal window handle being generated and the app quits. On my RiscStation, SA RPC, a friends Mico and even on an Omega, everything works as it should.
Any ideas on what could be causing it? |
| |
 [ Log in to reply ] |
| |
|
Adrian Lees |
Message #49908, posted by adrianl at 05:04, 20/1/2004, in reply to message #49785 |
Member
Posts: 1637
|
Are you sure your copy of RISC_OSLib is 32 bit clean? (I don't recall seeing an official 32-bit version. I've 'converted' my personalised copy - used by HDC - and it works with no problems, but it differs substantially from the original source release.)
I'd suggest you find an action that causes the program to fail every time and then insert trace statements into the source, to find out exactly where the error occurs.
Bear in mind that the illegal window handle may be a knock-on effect from an earlier problem that hasn't been detected & reported, eg. template handling, window creation...
Aside from the changes for 32-bit CPU mode, there's really not much else that's different between the Iyonix & RPC and relevant to this code. Dynamic areas, perhaps? Quite a lot of code has broken because the Iyonix will allocate areas at top-bit-set addresses. |
| |
[ Log in to reply ] |
| |
|
Adrian Lees |
Message #50210, posted by adrianl at 04:36, 25/1/2004, in reply to message #49908 |
Member
Posts: 1637
|
Are you sure your copy of RISC_OSLib is 32 bit clean? (I don't recall seeing an official 32-bit version.) *Sheepish grin*
There's a 32-bit OK build of RISC_OSLib on my hard drive and it must've come with the Castle tools!
I think it must be what psychologists call denial, but at least the nightmares about ROLib have stopped now  |
| |
[ Log in to reply ] |
| |
|
Paul F. Johnson |
Message #50544, posted by pfj at 21:20, 31/1/2004, in reply to message #50210 |
Member
Posts: 54
|
There's a 32-bit OK build of RISC_OSLib on my hard drive and it must've come with the Castle tools!
I think it must be what psychologists call denial, but at least the nightmares about ROLib have stopped now The version is 32 bit clean (well, druck's program says it is). The crash still continues. It could be that a sys call is returning a -ve number where 0 is the lowest allowed or -ve > the -ve it's allow (eg -2 when -1 is the lowest allowed)
It is annoying |
| |
[ Log in to reply ] |
| |
|
Acorn Arcade forums: Programming: OS5 Window problem |
