Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Another angle on this....

Re: Another angle on this....

From: Heinz Kiosk <no.spam_at_ntlworld.com>
Date: Wed, 20 Feb 2002 14:19:47 -0000
Message-ID: <yiOc8.120487$H37.15682824@news2-win.server.ntlworld.com>


> ----------------------------------------------
> 2. BFD = Big ..... Deal
> ----------------------------------------------
Further to previous response, I suspect that a bug in dealing with CLOBS when doing parameterised ODBC inserts is likely to be associated with this. I pass in fully quoted correct-case columnnames (automatically, as I can't predict what might become a reserved word in a future version of any platform that my code might be targeted against). However the internal machinations of Oracle ODBC/CLOB behaviour (out of my control) generate a supplementary internal statement with an *unquoted* reference to the column. Bad news if you've got CLOB column names like "COMMENT" in your app.

At first I thought, "the idiot who wrote that bit forgot that he should quote the name, just in case", but of course that part of the driver *can't* automatically quote the name, because it doesn't know if the case is right or not. Other users will write unquoted column names of random case. So it has no *trivial* way at that internal quote-stripped code level of knowing whether Comment refers to "Comment" or COMMENT. Someone at a higher level of dealing with the call would have had to upper-case all the unquoted stuff before stripping the quotes from the quoted fields. Easier to let the insert call crash and hope that no-one important notices.

Its only a surmise, but it feels right....still BFD? I guess so if you don't care about being able to do inserts of CLOBS whose names happen to something weird for a CLOB like "COMMENT". I haven't checked if non-parameterised inserts have the same behaviour, because I don't want to use them, they are very much slower in every platform that I've tried them.

With any luck you'll RTFM me with the "Automatically quote CLOB machinations" flag. I don't see it though, why not just fix the driver once the bug is spotted. Not an easy one for them though because its an inherent problem.

BTW its just occurred to me that because of the slow speed of retrieval allied with decent untuned write speeds of around 1k records/second I am in the bizarre (to me) situation that my routines write new records to Oracle tables about 10 times faster than they can retrieve them. Strange.

Regards Received on Wed Feb 20 2002 - 08:19:47 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US