Hello Peter!

In a message to Doug Glenn  Peter Fitzsimmons wrote:

PF> Because of a difference in interpretation of the DPMI spec,  any Borland
PF> product (or compiled program) has to have the OS/2 "Dos
Setting" for DPMI
PF> changed from "Auto" to "Enabled".

     And this is where the problem arises. Because it is a
    beta, I expect to find buglets, and the causes. Thus, when
    encountering a problem, I usually go through several sequences
    in attempting to track it down, usually 3-4, and in some cases
    where there are more variables, as many as it takes.

    I have tried AUTO and ENABLE in the settings along with the /UMB
    and DOS HIGH statements, Both in the Config.sys file and the DOS
    settings box. Although I don't consider myself an expert with
    OS/2 by any means, (besides, I am of the mind that once you
    claim to an expert, either there is no more to learn in the
    field, or you'vs stopped learning )  I am familiar enough
    with it to be comfortable making whatever changes I feel might
    be needed to track down a problem. Troubleshooting techniques,
    once learned, work pretty good for any environment.... providing
    you follow logical steps in isolating the problem and ensure
    your familiar enough with the environment to be able to figure
    out if you caused the problem .

PF> I think the problem arises because the programs in question do not do a
PF> "dpmi presence check" before the first allocation.

    They had written a program that merely allocates and deallocates
    a section of memory as a test for me to use. Again, this one
    failed with the same error message the previous version did. Now
    one thing that does concern me as being consistant, is that it
    reports (025) unable to find loader besides the SYS3175 error.

    This seems to me, to point to the loader provided (ie:RTM.EXE
    and DPMI16BI.OVL) rather than their coding implementation.
    However without some conclusive proof, ie: an experianced OS/2
    programmer that is familiar with the Borland products and coding
    for OS/2 or under OS/2 that has run across a simular problem and
    fix, then I will be unable to prove that it is OS/2 that is the
    problem here.... and that I would like to disprove.

    However, so far I have not gotten a definate answer from here,
    and will drop the thread as it is not producing an answer
    (either good or bad), and it is a borderline topic for the
    conference.

    I have not contacted Borland about the problem yet because I do
    not own any of their products, however that may be what I have
    to do in the final analysis in order to get to the bottom of it.

    Thanks for your patience and input!

 * WCE 2.01á1/2227 * In a VERY weak moment, I started a BBS.

--- WILDMAIL!/WC v4.10c