Andrew Grillet wrote in a message to Mike Bilow:

 AG> I have some stuff that uses the parallel port to emulate
 AG> IEEE488 and other such like things. This needs to CLI to get
 AG> H/W timing right.

 AG> I run it in a DOS box, cos I havn't had time to port it to
 AG> OS/2 yet.

 MB> You have no business doing things like this under OS/2.  You can't
 MB> even grab a hardware interrupt from a DOS box, but can only get it
 MB> reflected.  This changes a code path that could be a few bytes long to
 MB> one that is hundreds of bytes long, and this blows away your timing
 MB> constraints anyway.  Properly, you should do this with a device
 MB> driver.

 AG> Come on .. One of the big advantages of OS/2 is that it DOES
 AG> run DOS software.

OS/2 runs most DOS software.  It will not run DOS software that is
critically dependent upon external real-time constraints.

 AG> Sure it would be better to port the app to
 AG> OS/2, and yes, I know it should be a device driver, but
 AG> there is one major stumbling block ... AFAIK there are only
 AG> 24 hours in a day. If you can supply me another 8, I will
 AG> have the thing ported in a trice.

 AG> I am all ready to port it ... but Im spending 8 hrs a day
 AG> porting video DDs. The printer is low priority. The SCSI one
 AG> will be started as soon as video is done.

I think you are pushing OS/2 past its design limitations.
 
-- Mike


---