🚀 go-pugleaf

huver@amgraf.UUCP (Huver) writes:

>In article <1990Nov18.030410.7819@hoss.unl.edu>, Phil Dietz wrote:
>NOTICE, however, in Lydiatt's setDate() function he used:

>   arg[1] = (ULONG)IoErr(); /* lock on parent director set by DeviceProc() */

>which is incorrect.  ParentDir() should be used for this purpose.

>A similar function that I use with Manx 3.6/5.0 (under WB 1.3 or 1.3.2),
>modified slightly from original PDmake posting, is attached below.

On the contrary, using (ULONG)IoErr() is quite correct, and using
ParentDir() won't work in most cases.  See, DeviceProc() is a somewhat
magical function, at the heart of AmigaDOS.  You call it with a pointer
to a volume:path/filename - it returns in D0 the address of the message
port of the associated filesystem, and in IoErr() the parent directory
lock relative to the volume:path/filename you passed.  This is fully
documented in the Tech Ref Man.

Calling ParentDir() will only work correctly if there is no
volume:path/, or you strip that off before calling ACTION_SET_DATE. 

It is my understanding that if I pass the name "a/b", the appropriate
parent directory lock to pass is the current directory lock, whereas if
I pass "a:b/c", the appropriate lock to pass is a lock on a: (obtained
from the device list, I think).  I've looked through a few filesystems
and this seems to be correct.

>-huver  ...!uunet!amgraf!huver  -or-  huver%amgraf@uunet.uu.net

  Gregory Kritsch                          | University of Waterloo
    Fido:  1:221/208.11110  [1:163/109.30] | 1A Computer Engineering
    OCUG:  ggk@tirith.ocug.on.ca           |----------------------------
    UUCP:  ggk@tirith.UUCP                 | The University doesn't get
           ...!watmath!xenitec!tirith!ggk  | a chance to censor me!