Continuing the theme of how wonderfully 'stable' 4.6 is. Just completed a conversion for a client from 1.5 - 4.6 onto NT. A screen had been written with 2 LS's. The first had a file in Change mode, the second (a paging screen) had the same file in change mode. Both file accesses were accessing the same record, and the net affect was to blank out data in the record apart from the key!! Now that is bad enough but the fact that Pro-IV did not issue a record lock is very concerning, it obviously has MAJOR data integrity concerns. Obviously the function design was very strange and in fact the change mode in the primary LS was unneccessary but that is not the point.
major bug
Started by
Guest_gareth thomas_*
, Dec 17 1999 10:12 PM
1 reply to this topic
#2
Posted 17 December 1999 - 10:20 PM
The record locking behaviour that you describe is the normal behaviour under UNIX, but under VMS it DOES issue a record lock! This is of itself a significant problem as some parts of SuperLayer lock under VMS because of this.
You would expect to get the change passed through in your case, but there are worse problems if you can get multiple locks on the same record within nested global functions as the changes are not passed through (because of scoping)
At this time my advice to all and sundry would be to stick with 4.0 for production systems and ignore 4.6 until it becomes a bit more stable.
Cheers
Dan Shannon
You would expect to get the change passed through in your case, but there are worse problems if you can get multiple locks on the same record within nested global functions as the changes are not passed through (because of scoping)
At this time my advice to all and sundry would be to stick with 4.0 for production systems and ignore 4.6 until it becomes a bit more stable.
Cheers
Dan Shannon
Reply to this topic
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users