How to identify the lock acquired by another user on a record in progress openedge - progress-4gl

Question

User A has read first record from a customer table. User B wishes to know the lock which is held by A on the record. How will B identify the lock held by User A programmatically.

Answer 1

You can use LOCKED function:
Returns a TRUE value if a record is not available to a prior FIND . .
. NO-WAIT statement because another user has locked a record.
Example of the documentation:
REPEAT:
PROMPT-FOR customer.cust-num.
FIND customer USING customer.cust-num NO-ERROR NO-WAIT.
IF NOT AVAILABLE customer THEN DO:
IF LOCKED customer
THEN MESSAGE "Customer record is locked".
ELSE MESSAGE "Customer record was not found".
NEXT.
END.
DISPLAY cust-num name city state.
END.
Other example:
FIND FIRST table
WHERE table.c1 = "foo"
EXCLUSIVE-LOCK NO-ERROR NO-WAIT.
IF LOCKED(table)
THEN DO:
/* The record is blocked by another user */
END.
ELSE DO:
IF AVAILABLE table
THEN DO:
/* The record can be modified */
END.
END.

Answer 2

You can use the built in _Lock Virtual System Table.
This is a basic example:
DEFINE TEMP-TABLE ttLock
FIELD LockId LIKE _Lock._Lock-Id
FIELD LockUsr LIKE _Lock._Lock-Usr
FIELD LockName LIKE _Lock._Lock-Name
FIELD LockTable LIKE _Lock._Lock-Table
FIELD LockFlags LIKE _Lock._Lock-flags
INDEX LockIdx IS PRIMARY UNIQUE LockId.
FOR EACH _Lock NO-LOCK:
IF _Lock._Lock-Usr = ? THEN NEXT .
CREATE ttLock.
ASSIGN
LockId = _Lock._Lock-Id
LockUsr = _Lock._Lock-Usr
LockName = _Lock._Lock-Name
LockTable = _Lock._Lock-Table
lockFlags = _Lock._Lock-flags.
END.
FOR EACH ttlock:
FIND _Trans NO-LOCK WHERE _Trans._Trans-Usrnum = ttLock.LockUsr NO-ERROR.
FIND _File NO-LOCK WHERE _File-Number = ttLock.LockTable.
MESSAGE
"Transaction Id:~t" (IF AVAILABLE _Trans THEN _Trans._Trans-Id ELSE ?) "~n"
"User Number:~t" ttLock.LockUsr "~n"
"User Name~t" ttLock.LockName "~n"
"Table Number:~t" ttLock.LockTable "~n"
"Table Name:~t" _File-Name "~n"
"Flags:~t" ttLock.LockFlags
VIEW-AS ALERT-BOX INFO BUTTONS OK.
END.
(Taken directly from this entry in the knowledgebase.)
In the Database Administration guide you can see (page 848) you can see what the flags in _Lock.LockFlags stand for.
Flags for the lock—the flags specify a share
lock (S), exclusive lock (X), a lock upgraded
from share to exclusive(U), a lock in limbo (L),
a queued lock(Q), a lock kept across
transaction end boundary (K), a lock is part of
a JTAtransaction (J), a lock is in create mode
for JTA (C), or a lockwait timeout has expired
on this queued lock (E)
Querying the _Lock table might be something you want to do in test environments only. Depending on your systems size there might be LOT of data there. Also, use VSTs for READ-ONLY operations!

Answer 3

Querying _LOCK in older releases is OK, but you have to use the appropriate code for the respective version:
_LOCK always has the full number of entries given in -L, regardless how many locks currently exist.
In pre-11.4 releases the fields are not indexed, but all used locks are at the beginning of the table, so you can use
IF _Lock._Lock-Usr = ? THEN LEAVE.
in the for each loop (_Lock._Lock-Name = ? is also fine). See http://knowledgebase.progress.com/articles/Article/P161995
In 11.4 and 11.5 the populated entries are no longer at the beginning so the old code will give wrong results (see http://knowledgebase.progress.com/articles/Article/000056304, this is fixed in 11.5.1). Fortunately scanning the lock table is now much faster so you can use
FOR EACH _Lock NO-LOCK WHERE _Lock-Recid <> ?:
mentioned in the same article. Technically this is not implemented with indices. (An index wouldn't work with the <> operator.)
In 11.5 and 11.6 both variants should work, but the newer variant with a where phrase should be faster.