Tim

I'm about to change the structure of a table to reduce the duplication of data. Its emails / news (ie its TMaNs replacement) and currently I hold the data in both encoded and decoded forms. I'm going to move to holding just the encoded data and decode as necessary. On the first decode I'll have to work out if its plain text, mime encoded or UUencoded and if there are attachments, I'll also set a field in the record as a flag as to which it is. Subsequently I can just check the flag and apply the right mechanism. Unfortunately I can't see any way of doing this for full text indexing since I can't see any way of getting back to the record that's being indexed.

Any Ideas or useful suggestions (apart from wait for ElevateDB)?


Roy Lambert

I think I've figured out a way - set a couple of variables in a beforepost event. Can I have your advice on wether they should be private to the unit in which the fti procedure is declared, global or what?

Roy Lambert

Roy,

<< I think I've figured out a way - set a couple of variables in a
beforepost event. Can I have your advice on wether they should be private to
the unit in which the fti procedure is declared, global or what?  >>

For local access, you should be able to get away with that.  However, at the
very least make sure that they are threadvars so that multiple threads don't
screw each other up.

One other thing you may want to consider is a separate field for UU-encoded
messages vs. mime-encoded messages.  It's not exactly pretty, but it will
work.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim


I'd need about a slew of  fields cos it could be plain text, uu encoded, mime encoded, rtf, ansi or html (I think that's 9). I do hope ElevateDB will have a couple more bells and whistles in this area - please its so useful but I keep running into bits like this.

Roy Lambert

Roy,

<< I'd need about a slew of fields cos it could be plain text, uu encoded,
mime encoded, rtf, ansi or html (I think that's 9). I do hope ElevateDB will
have a couple more bells and whistles in this area - please its so useful
but I keep running into bits like this. >>

Yeah, but you have to remember that the full text indexing filtering is done
in an abstract way, and isn't always tied to a specific record.  However,
I'll see what I can do.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim


I know - I've had to scrap my idea because I suddenly realised I had no idea what would happen with sql operations - I could have done some experimentation but I've decided to analyse the data as it arrives so I'm currently working on making that as fast as I can.

Thanks for your input as usual.

Roy Lambert