Re: Puzzeling error with package

William Robertson <williamr2019_at_googlemail.com> writes:

> On Oct 31, 10:43 pm, Tim X <t..._at_nospam.dev.null> wrote:
>> joel garry <joel-ga..._at_home.com> writes:
>> > On Oct 30, 6:13 pm, Tim X <t..._at_nospam.dev.null> wrote:
>>
>> >> thanks Joel. Thats close to what the DBAs and I were beginning to arrive
>> >> at. thanks for the reference. I'm going to look at metalink2 on the
>> >> weekend.
>>
>> > On reflection, I'm wondering if this might happen if you simply ran
>> > the reload procedure twice.  You wouldn't see the bad log if you
>> > always use the same name, maybe append the process id to it...
>>
>> thats possible. Good idea to append the pid to the log name.
>>
>> Still doesn't make sense to me - I remain puzzled over how the variable
>> can both be defined, but have a null value when its definition includes
>> initialisation and it is defined as a constant. Even if it was some sort
>> of caching issue, or something odd on the client/sqlplus side, as it is
>> a constant varchar2, it still shouldn't have a null value.
>>
>> Part of me wishes it would happen again just so that I would have a
>> chance to track it down. . Just don't like it when my code does
>> something unexpected that I can't explain. With my luck, it will do it
>> again at next weeks demonstration!
>>
>> thanks,
>>
>> Tim
>>
>> --
>> tcross (at) rapttech dot com dot au
>
> Also looks similar to this, which turned out to be a PL/SQL compiler
> bug:
> http://forums.oracle.com/forums/thread.jspa?messageID=2708266
> (note due to a bug in the current OTN forum software, '<>' displays as
> a blank, so Steve's problem occurs with the statement "Exit when x <>
> y;")

Thanks William, thats an interesting reference.

Tim

-- 
tcross (at) rapttech dot com dot au