Path: utzoo!utgpu!water!watmath!clyde!rutgers!lll-lcc!ames!hao!husc6!mit-eddie!ll-xn!adelie!infinet!ulowell!page
From: page@ulowell.cs.ulowell.edu (Bob Page)
Newsgroups: comp.sys.amiga
Subject: Re: Disks
Message-ID: <2149@ulowell.cs.ulowell.edu>
Date: 18 Dec 87 18:01:43 GMT
References: <3186@batcomputer.tn.cornell.edu> <5985@cisunx.UUCP>
Reply-To: page@swan.ulowell.edu (Bob Page)
Organization: University of Lowell, Computer Science Dept.
Lines: 82

>Check out FFormat, a program by Andy Rachmat (sp?).  It formats a disk
>in half the time it takes AmigaDOS format.  

I wonder about this.  I ran it with verify mode on, like FORMAT does,
and didn't see such a dramatic increase in speed.

What really concerned me is FFormat writes all zeros to the sectors,
and then checks the zero'd sectors.  Seems to me writing a varying
pattern to the disk is a better test of verification.  FORMAT writes
like:

UBYTE word;
register i2, i3, sec;
UBYTE *val2 = "Oo";	/* second char of pattern */
UBYTE *val3 = "SsWw";	/* third char of pattern */

  for (sec=1;sec<4;sec++) {	/* do four sectors this way */
     for(word=0; word<BLOCKSIZE; word++)
        buf[word] = sprintf("D%c%c%d", val2[i2], val3[i3], word)
     Write_to_disk(unit, buf);
  }

This is from memory off the top of my head, as I don't have an Amiga
handy and it was a long time ago I studied what FORMAT did.  The
longwords in the sectors look like DOS%d, where %d is the longword
offset into the sector.  That's why sector 0, longword zero says
"DOS\0" ... because that's what FORMAT writes there.  After four
sectors like this, it picks a different pattern, like DOW%d or DoS%d
or something, and writes it for four sectors.

I don't claim this is the best test for checking failures; a extensive
test would write all zeros, then all ones, then alternate zeros and
ones (hex AA and 55), testing after each pass.  But at least this
verification tests some data, over characters 0-127 of the ASCII
character set, rather than FFormat's zeros.

I'll stick with Format, thank you.

HOWEVER, if what you really want to do is quickly zap a disk of all
its files, you just want to unlink all the files from the hashlist in
the root block.  Doug Walker's DiskWipe does this.  You also have to
rebuild the disk's block allocation bitmap, and I don't remember if
DiskWipe does. [I guess it must, else the disk validator would get
re-run and rebuild it anyway]

======
Somebody wanted to rescue data from a file when it was locked for
writing by DOS and the program got a Task Held requester.

You can do this with Sectorama if you know the pathname to the file;
start it up, Find the hash value for the first directory name in the
path, Jump to that block (possibly follow the hash chain to the one
you want), and keep doing this until you're at the directory and file
you were looking for.  Then use the Data and Write commands to step
through the data blocks, saving to some file.  You won't get any
buffers that weren't flushed to disk, but you'll rescue some of your
data, anyway.

It easy to cook up a simple RESCUE command to do all this for you.

=====

>There is a program called GOMF (Get Outta My Face.  I'm not kidding.)

GOMF1.0 has some problems.  I had a bunch of windows open on a custom
screen, so I went to WHAP them.  I couldn't see I had WHAP'ed all the
windows and the screen (and it didn't tell me), so the next time I
WHAP'ed a window it destroyed my editor, which was the active window
on the Workbench screen.  Since I'd already LOST some work, I just
kept WHAPping things - my CLIs, clock, etc, until there was nothing
left but the WB screen.  I WHAPped it and ... GOMF went GURU #3.

=====

All the aformentioned programs (except for RESCUE) are publicly
available; I got them from People/Link and the Fish Disks.

..Bob
-- 
Bob Page, U of Lowell CS Dept.  page@ulowell.edu  ulowell!page
"I've never liked reality all that much, but I haven't found a
better solution."		--Dave Haynie, Commodore-Amiga