archival cd-r - really true?

[Doc Shipley]

Chuck Guzis wrote:
> through at close to 100%.  I've lost CD-Rs simply by dropping them and
> having them roll under a desk (dirt on the floor will do a number on a CD-R
> (or CD) if the thing gets dragged across the floor.

   I found an interesting tidbit a couple of weeks ago:

http://www.burningissues.net/how_to/scratchrepair/scratchrepair.htm

   and

http://www.mcgee-flutes.com/scratches.html

   Both recommend plain ol' Brasso as the best available CD repair.  The 
burningissues page even used some pretty good metrics for comparison.


   (Chuck, I expect you know all the following; I'm just tossing it in 
here 'cause I'm lazy)

   I haven't done as much with damaged data CDs as with "odd" format 
CDs, but I've found that Linux and NetBSD have *far* fewer issues with 
burned media vs reader issues than Windows or MacOS.  I regularly use 
disks burned in Linux (on *cheap* media) in RRD42s on VAX, old 68K Macs 
with 300i drives, etc.

   I think a lot of the "CD reader" issues are really "CD writing 
software" issues.  A lot of CD-copy software will insist that all the 
world's ISO9660 unlss you beat it up first.  Even dd won't produce a 
bootable image in some formats unless you tweak the blocksize.  (AIX 
v.<any> and IRIX v6.5 come to mind)  Do some research on the format 
you're copying and the software you're using, life gets much smoother.

   On data CDs, setting dd to a stupidly small blocksize can help 
recover files off a damaged or funky disk.  I've gotten all the data off 
a CD using "dd ... bs=1" that I couldn't get trying to read the files 
directly.  It took a while.  ;)


	Doc