| TIP: Click on subject to list as thread! | ANSI |
| echo: | |
|---|---|
| to: | |
| from: | |
| date: | |
| subject: | viogetbuf |
>> PE> What I am wondering is whether >> PE> EVERY 32-bit address I can allocate can be converted >> PE> into a 16:16. If it >> Yes, you can convert any address from a 0:32 (actually, if >> you want to be pedantic, this should be called 16:32) to a >> 16:16 and vice versa. dn> Hmm, look again. You can't fit 48 bits into 32 no matter how big a hammer dn> you might use. :-) When you use DosAllocMem(), you can specifically dn> request 'tiled' memory (memory which can be converted to 16:16) or not. Ok, you're right. In the full 4 gigabyte address space, not all bytes area addressable by a 16:16 pointer. However, OS/2 currently only allocates 512 meg to each process, so all bytes are addressable by both 16:16 and 0:32 addresses. According to what I've read, all memory requests are forced to be tiled under the current version of OS/2. Obviously this may change and if you need compatability with 16:16 code you should expicitly request tiled memory. dn> It is actually 0:32 since separate segments aren't used in flat model. If dn> the OS/2 kernel allowed modification of segment registers in 32-bit code dn> segments (and it does, but for ring 0 code only), then it might arguably dn> be called 16:32. Each process has one segement. It's just 4 gig in size :-) >> These sorts of problems will keep occuring until IBM pulls >> their finger out and rewrites these routines as 32 bit code. dn> According to an IBM enginerr I exchanged some email with recently, IBM dn> already have 32-bit versions of their subsystems, but from what he said I dn> doubt that they'll ever be seen out in released code. The general dn> approach is that (1) IBM isn't committed to Vio applications, (2) that dn> IBM aren't guaranteeing forward compatibility with ports of OS/2 to other dn> architectures and (3) IBM doesn't want to 'waste' resources on bug fixes dn> and enhancements to an area of the operating system in which they are not dn> particularly interested. I guess they've also got to be careful not to break the existing 16 bit applications. dn> In other words, it's not broke and it doesn't need fixing, and especially dn> doesn't need enhancing. dn> Pity. I quite like text mode apps myself, but I came to face the music dn> on hearing this. I'll still continue to use (and even produce) text mode dn> apps in spite of it, but I've already shifted into GUI development. The steep learning curve developing GUI applications is a pain. It's almost enough to take the fun out of programming :-( >> And you thought that we'd left all this stuffing around >> behind us when we went OS/2 :-( dn> You'd rather use software interrupts and an API like Windows? :-) No, but I do have this soft spot for punched card output :-) Paul --- GoldED 2.40* Origin: It's life Jim, but not as we know it (3:711/934.1) SEEN-BY: 711/809 934 @PATH: 711/934 |
|
| SOURCE: echomail via fidonet.ozzmosis.com | |
Email questions or comments to sysop@ipingthereforeiam.com
All parts of this website painstakingly hand-crafted in the U.S.A.!
IPTIA BBS/MUD/Terminal/Game Server List, © 2025 IPTIA Consulting™.