Thanks,
I've already found a few. I tested them and you're correct. I saw it
happen. We're still
looking for more of those.
Pam
Stevenson, Charles wrote:
I know you already said it's ok, but I'd double check for READs where
READUs belong.
Especially since you've described this as having a Pick heritage from a
time when some Pick implementations and programmers did not handle
locking well.
One user's changes can easily wipe out another's.
WRITE will make use of the lock table even when the program does not ask
for it explicitly.
Consider 2 programs updating the same record, one using READU, the other
without explicit locking, where the chronological chain of events is
this:
0. --- on disk Rec 123: 'A~B~C~Moby Dick' ---
('~' is an attribute mark)
User Jack User Jill
========= =========
Wants to Wants to
change Moby Dick change A~B~C
to Tom Sawyer to X~Y~Z
1. READU REC FROM F,'123'
- sets lock
- gets 'A~B~C~Moby Dick'
2. REC<1> = "X"
REC<2> = "Y"
REC<3> = "Z"
in private memory:
(X~Y~Z~Moby Dick)
3. ---( on disk still as 'A~B~C~Moby Dick' )---
4. READ REC FROM F,'123'
(ignores lock, sets none.)
gets 'A~B~C~Moby Dick'
5. REC<4> = 'Tom Sawyer'
In private momory :
(A~B~C~Tom Sawyer)
6. WRITE REC TO F, '123'
7. behind the scenes UV check for
& ends up waiting for Jill to
release the lock...waiting...
...waiting...waiting...
8. ---( on disk still 'A~B~C~Moby Dick' )---
9. WRITE REC TO F,'123'
10. ---( on disk rec 123: 'X~Y~Z~Moby Dick' )---
11. readu lock is released.
12. UV *implicitly* 12. CRT 'Jill, success: A~B~C was
sets the lock on changed to X~Y~Z. Congrats.'
Jack's behalf!
Jack's version of
the rec is written,
readu lock released.
13. ---( on disk rec 123: 'A~B~C~Tom Sawyer' )---
Time passes
14. Jill calls Pam, "Hey! Where are my changes?
Computer said I changed A~B~C to X~Y~Z
but it's A~B~C again!?!? I think we need MSSQL."
Other variations on this theme will yield similar corruption if READUs
are not used consistently, exclusively, and explicitly whenever updates
are performed.
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
--
Pamela J Robbins New Bolton Center
Senior Programmer Analyst School of Veterinary Medicine
[EMAIL PROTECTED] University of Pennsylvania
610-925-6438
-------
u2-users mailing list
u2-users@listserver.u2ug.org
To unsubscribe please visit http://listserver.u2ug.org/
