Re: ? disk writes actually done or still just in the cache ?

[bill@cambridge.apple.com (Bill St. Clair)]

>X400-Received: by /PRMD=inria/ADMD=atlas/C=FR/;	Relayed; 19 Jun 92 08:37:23+0200
>X400-Received: by /PRMD=CNET/ADMD=ATLAS/C=FR/;	Relayed; 19 Jun 92 08:36:00 GMT
>Date: 19 Jun 92 08:36:00 GMT
>From: Ranson <ranson@LANNION.cnet.fr>
>To: bill@cambridge.apple.com, ranson@lannion.cnet.fr
>Subject: Re:  ? disk writes actually done or still just in the cache ?
>Cc: info-mcl@cambridge.apple.com
>
>I think both force-output and finish-output should use _FlushVol. I think that
>using FlushFile at all is dangerous. If the machine is reset for any reason
>after the _FlushFile but before any _FlushVol, the disk may be corrupted.
>     Daniel.

David Moon already called me on it, though for a slightly different reason.
I have decided that the patch will change the current #_FlushFile to
#_FlushVol.

>Date: Thu, 18 Jun 92 11:07:22 EDT
>From: moon (David A. Moon)
>To: bill@cambridge.apple.com (Bill St. Clair)
>Subject: Re:  ? disk writes actually done or still just in the cache ?
>
...
>> 
>> In 2.0b1p3 and later, FORCE-OUTPUT on an output file stream calls
>> #_FlushFile. CLOSE calls #_FlushVol. The Inside Macintosh documentation
>> for #_FlushFile says that "Some information stored on the volume won't
>> be correct until PBFlushVol is called." Seems to me that FORCE-OUTPUT
>> should do #_FlushFile, which it already does (in 2.0b1p3. 2.0b1 without
>> patch 3 tried to do #_FlushVol, but neglected to put the volume number
>> in the right place), and FINISH-OUTPUT should do #_FlushVol.
>
>That seems wrong, since FORCE-OUTPUT and FINISH-OUTPUT are supposed to do
>the same thing except that the latter doesn't return until the I/O has
>completed while the former returns sooner if it feels like it.
>
...