I regularly miss messages from the SF mailing lists since 1-2 weeks. So I have to wade through the big traffic web interface to find out what's out. This way I found your reply to my message below:
> + if ((result & (IS_NETWORK | IS_DEVICE)) == IS_NETWORK) > { > int status; > if ((flags & (O_TRUNC | O_CREAT)) == O_CREAT)
I didn't apply this patch since it is wrong: H:NUL should give H:/NUL in truename which results in only the IS_DEVICE bit set (unless I missed a bug in truename).
EJECT uses "H:CON" and if I don't apply this patch, it doesn't work. So the patch may be wrong but it fixes the bug, whereas leaving it without it looks better theoretically but in practice fails. Which is better? Maybe a better patch is needed? In any case, this bug must not be ignored.
IS_NETWORK|IS_DEVICE both set is for things such as H:\DOS\NUL which really should go via the network redirector to check if H:\DOS is valid!
Please see my explanation I sent Friday which showed up a little late for me (but I sent it privately too!).
Sorry, I didn't receive anything, neither from the list, nor from you. I will send you a personal message from my private address to which you can reply and re-send me your message. Hope it works.
try to investigate if anything weird happens with the high parts of the eax, ebx, ... registers. Borland may rely on them in the multiplication/division routines (at least it does in turbo pascal; it might have done the same for borland C too)
OK. Thanks for the suggestion. Soon or latter, shall catch it! ;-)
Lucho
------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ Freedos-kernel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/freedos-kernel
