Hi,
On Sun, Sep 21, 2003 at 05:15:18PM +0200, Michael Stefaniuc wrote:
Alexandre, please use following revised changelog for the patch: Changelog:
- silence the FIXME's produced by a check for dos4gw or compatible dos32 extender.
Maybe it's actually a good idea to change the code to return "support", since we already have DPMI support (it might not be 100% perfect, but it's almost 100% working, IIRC).
- silence the FIXME produced by an install check of MVSOUND.SYS
That's one FIXME less in exchange for one additional code path, which would have to be done for every utility on earth that uses int 0x2f for install checks (since they all use different values for their checks). Maybe this shouldn't be done like that, who knows...
Andreas Mohr
Hello,
On Sun, Sep 21, 2003 at 05:44:23PM +0200, Andreas Mohr wrote:
On Sun, Sep 21, 2003 at 05:15:18PM +0200, Michael Stefaniuc wrote:
Alexandre, please use following revised changelog for the patch: Changelog:
- silence the FIXME's produced by a check for dos4gw or compatible dos32 extender.
Maybe it's actually a good idea to change the code to return "support", since we already have DPMI support (it might not be 100% perfect, but it's almost 100% working, IIRC).
You are confusing something: the int21 0xff functions are specific to dos4gw and compatibles and have nothing to do with DPMI. The DPMI installation check is int2f AX=1687h and that one already returns "success".
- silence the FIXME produced by an install check of MVSOUND.SYS
That's one FIXME less in exchange for one additional code path, which would have to be done for every utility on earth that uses int 0x2f for install checks (since they all use different values for their checks). Maybe this shouldn't be done like that, who knows...
Yeah, i've thought about that too. There are a lot of interrupt functions in Ralf Brown's interrupt list that are provided by third party (not DOS or Windows) drivers/programs. Adding code to explicitly ignore them would add a lot of bloat. There are two ways to do it: a.) like we do now: add the ignore code if a program falsly triggers an ERR b.) like DOS and the BIOS are doing it: if the interrupt handler dosn't know about the function then do nothing and return. But for this we would need to go through all interrupt handlers and add FIXME's for all the functions that still need to be implemented by winedos.
The first approach was easier in the beginning, migrating to the second method would be yet another janitorial task. But this is not for me to decide, Jukka/Alexandre are the people to make the call.
bye michael
-
Andreas Mohr -
Michael Stefaniuc