Subject | Re: Potential computed by and UDF issue |
---|---|
Author | robschuff |
Post date | 2005-02-09T17:01:04Z |
Ahhhh...I neglected to mention that the table **DID NOT** restore and
there was no error reported in the verbose output until the permission
on the table was restored. I have tried this on three different
installations of the firebird version I mentioned with identical results.
I orginally discovered this using FreeUDFLib and my original syntax
was not a problem. I wanted to check with a UDF that is distributed
with FB, so I used it and only tested on the empty database hence I
didn't catch the incorrect usage of the UDF. But as Helen has noted
this issue should not be relevant with this example.
Though it seems to not be a problem with FB 1.5.2, I think it still
should be investigated as a potential **SERIOUS** bug. Since gbak
(and gfix) reports no errors during backup one potentially have a
bunch of useless backups.
thanks!
Rob
--- In firebird-support@yahoogroups.com, Helen Borrie <helebor@t...>
wrote:
there was no error reported in the verbose output until the permission
on the table was restored. I have tried this on three different
installations of the firebird version I mentioned with identical results.
I orginally discovered this using FreeUDFLib and my original syntax
was not a problem. I wanted to check with a UDF that is distributed
with FB, so I used it and only tested on the empty database hence I
didn't catch the incorrect usage of the UDF. But as Helen has noted
this issue should not be relevant with this example.
Though it seems to not be a problem with FB 1.5.2, I think it still
should be investigated as a potential **SERIOUS** bug. Since gbak
(and gfix) reports no errors during backup one potentially have a
bunch of useless backups.
thanks!
Rob
--- In firebird-support@yahoogroups.com, Helen Borrie <helebor@t...>
wrote:
> At 07:42 PM 8/02/2005 +0500, you wrote:somewhere
>
> >Hello Rob,
> >
> >r> Does anyone care to take this question on please? Is there
> >r> else I should report potential bugs?UDFs, it's
> >AFAIK, it is a known problem: restoring of UDFs occures after
> >restoring of tables, so computed field(s) cannot know your UDFs at
> >restore time. As I've read earlier, FB2 will fix the problem (I may be
> >wrong).
>
> No; it was touted as a possible problem. But, with respect to
> a non-existent problem. It WILL be a problem if you try to definecomputed
> fields based on calls to UDFs that are not declared in the database,or are
> incorrectly declared (which is not the problem that was reportedhere); or
> if you base computed fields on references to tables, views or storedissue,
> procedures that haven't yet been restored. This latter *is* a known
> since restore restores each group of objects in alphabetical order, notUDF
> creation order.
>
> Since the restore didn't fall over when creating the table (with the
> reference) but when trying to restore permissions, Rob needs tolook at
> the permissions and find the error there that is causing the restoreto fail.
>
> ./heLen