[kernel] Text/method fields

Erik Mueller-Harder lists at praxisworks.com
Fri Aug 8 11:22:26 CDT 2003


Ruslan Zasukhin wrote:

>> 2.  and you can make a Text/method field that is based on
>>     VarChar fields (why would you want do this?  If you're
>>     concatenating field values.) -- works fine (with only
>>     limited testing);
> 
> For example you have concat of few VarChar, so size can be >64KB

Exactly.  I'm doing this and it works fine.

> 
>> 3.  you can NOT make a Text/method field that is based on
>>     another Text/method field.  Result seems invariably to be
>>     empty.  That's too bad, because I'd wanted to concatenate
>>     various other results into one large Text/method field for
>>     indexing purposes.  Sigh....
> 
> I think this is really not implemented, because of potentially
> unlimited size of RESULT VALUE.
> 
> How I can handle this?
> * Even one TEXT value can be few MB/GB
> * of course it is possible produce result to disk, so this is not
> trivial step.

Interesting points.  In my case, I'm trying to concatenate one Text/method field which typically will be in range of 4-5K (but could be as much as 250K occasionally) with another that is usually empty (but could be 1K or 2K) and a few VarChar fields (each of which is uaually only a few hundred bytes, but a couple of which could sometimes be as much as 64K).  I expect my final concatenated Text/method field usually to be only 5K or so, though it clearly could be as much as 400K occasionally.

I see the implementation problem, though.  I should be able to work around the limitation with more complex SQL inquiries.

Thanks for your help!

Erik


More information about the Valentina mailing list