Everything that RHD said is bang-on.
As Cringer says, that looks like an odd window name. Also, the colon after the window name looks like an error to me. It might be legal -- but it is certainly not normal.
Are you really planning to do translation? Or are you just sprinkling :U on strings for the joy of it?
Where do you expect FIND CURRENT to be looking for a current record? This snippet of code requires it to look outside the scope of the trigger so you are relying on something which is globally scoped and potentially introducing side effects by changing the LOCK status. So that's a problem.
You're not actually doing anything with "users" other than (possibly) changing the lock status and using the username field to look up a logRest record -- so refinding it by way of a borrowed buffer and then changing the lock status makes even less sense.
You suppress the possibility of an error but there is no test for success of your FIND CURRENT. If FIND CURRENT fails you will quietly go on to the next statement.
The FIND of logRest does not have a NO-ERROR. So the IF AVAILABLE test isn't going to be effective -- an error will be thrown before you get there.
If LogRest is actually unique then FIRST serves no purpose and confuses people.
If LogRest is not unique then what are you doing with the other members of the set?
Either way, the FIND LogRest does not specify a LOCK and is only weakly scoped. It is a default share-lock and if this buffer is being borrowed from a globally scoped buffer (with code written like this that seems likely) then you are at risk of running into or creating lock conflicts.
I would rewrite this as:
Code:
on "window-close" of this-procedure
do:
define buffer del_logRest for logRest.
if available users then
do for del_logRest transaction:
find del_logRest exclusive-lock where del_logRest.userName = users.userName no-error.
if available del_logRest then delete del_logRest.
end.
else
do:
/* some sort of appropriate error handling or notification */
end.
return.
end.
