Msg: 6646 *Conference*

07-01-96 13:37:41

From: RON WIESEN

To : ROBERT BENSON

Subj: REPLY TO MSG #6637 (RAMPAC DIAGNOSTIC)

Don't have to remember -- I documented it here Bob.  For full details, see
MSG#5540 which is the warning I put here some time ago.  The relevant part of
MSG#5540 that describes scenarios is repeated here below.
__________
For .BA and .CO files, NPL works fine.  For .DO files, NPL sets the stage for a
disastrous RAM100/TS-DOS conflict!

Even if NPL is not ever used again, a virus-like situation ensues the first 
time you use RAM100 or TS-DOS to load any .DO file into the Model-T that had 
been put on the media (RAMPAC or PDD) by NPL.  For clarity, let's call this .DO
file the "viral source" that eventually leads to a cold start of the Model-T, 
and call its media image the "infected media" to remind you of where it came 
from and where it and perhaps others like it still lurk.

At the instant of arrival, the "viral source" corrupts the File Directory which
the Model-T operating system maintains, and consequently corrupts some or all 
other files in memory at this time (all types .BA, .CO, and .DO).  You now have
a corrupt File Directory AND corrupt files.  Unfortunately, the corruption is 
not immediately evident to you.  If you copy corrupt files from the Model-T 
memory to other media (by means other than NPL), the good news is that you 
don't make more "infected media" but the bad news is that you aggravate or make
impossible the task of recovering your files later.

When you select a file name from the now corrupt File Directory shown at the 
main menu, typically the operating systems invokes a different file (e.g., you 
select REPORT.BA and SORT.CO is invoked).  If it invokes a corrupt .CO file, it
may just beep and present the main menu again because the corrupt header of the
.CO file is nonsense compared to HIMEM.  Things are confused!  You become 
confused.  At some point either you intentionally cold start the Model-T or it 
does so.

Now you must reload some files (and set DATE$/DAY$/TIME$ and so on).  During 
the reload, you might use an "infected media".  Note that by this time 
"infected media" may mean both the RAMPAC and one or more PDD diskettes!  Of 
course you think all media is fine.  If you load any "viral source" the 
nightmare starts all over again.  If you don't load any "viral source" all 
appears well and you (incorrectly) conclude that some unexplained disturbance 
occurred that's not likely to be repeated.
__________
Also see MSG#5543 and it's replacement legend MSG#6519.  These give a graphic
picture of what goes on.

The heart of the problem centers on the EOF byte for ASCII (.DO) files.  It's a
case of different handling of EOF by different sofware.  Hope this clarifies.