You may want to reconsider point #3. Yes, it will greatly increase the size
of your database but it will also greatly increase the responsiveness of
your searches. How many records are in the table and what's the average
count of words per blob? What's your current average time on a search?
Would you like it to be near instantaneous?

This is exactly the kind of thing the full text searching stuff was designed
for.

Jason Wharton

> -----Original Message-----
> From: Vahan Yoghoudjian [mailto:vahan@...]
> Sent: Monday, November 15, 2004 11:55 PM
> To: IBObjects@yahoogroups.com
> Subject: RE: [IBO] Searching into Formatted memo fields
>
>
>
> Thank you for your replies Helen and Jason
>
> I think point 1 is the most appropriate. I will try to
> dig into RichEdit
> codes to find out how rich text is converted to plain text like Helen
> suggested.
>
> Point 2 would be much space consumming knowing that I
> will be using this
> search on a stock item table that might reach to
> 20,000-30,000 records with
> some clients, but in case point 1 doesnt work I think this
> would be our best
> choice left
>
> -----Original Message-----
> From: Jason Wharton [mailto:jwharton@...]
> Sent: Tuesday, November 16, 2004 7:38 AM
>
>
> > I have a TIB_RichEdit connected to a memo field.
> >
> > I want to use a Query to search in this memo field, but
> > everytime I use
> > words like 'font', 'size', 'color' as search criterias I have
> > a result set
> > even if I don't have those words in the field value. Is there
> > a way I can
> > search into Formatted text by using a query by ignoring the
> > text format
> > tags?
>
> None that I know of. There are three routes you can take:
>
> 1) Write a blob filter and somehow invoke it via the text searching
> mechanism. (I have no idea how or if you can do this.)
>
> 2) Maintain a separate column that contains only the plain
> text version of
> the field and search on it in place of the rtf version.
>
> 3) Use the full text searching mechanisms available in IBO
> and pass it the
> plain text version of the rtf stuff so that it builds its
> indexes using only
> the text content.
>
>
> Jason Wharton