I have been asked for information about indexing symbolics.
I have always suggested that users do not index symbolics but there seems to be an instance where it really is necessary (or at least the easiest workaround).
I have never seen anything from RTI either yea or nea about this and so am asking here.
Does indexing symbolics work?
Do the indexes keep maintained?
What if it is a simply symbolic .. just appending a character to a field?
Thanks
[email protected] onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"
David Tod Sigafoos ~ SigSolutions
Phone: 971-570-2005
OS: Win2k sp2 (5.00.2195)
OI: 4.1
DSig,
You have to be careful with symbolics getting their values from other tables.
- Oystein -
Oystein,
I will put that on the list. Just getting a "do/don't" list for the client.
Thanks for the quick response
[email protected] onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"
David Tod Sigafoos ~ SigSolutions
Phone: 971-570-2005
OS: Win2k sp2 (5.00.2195)
OI: 4.1
I believe this is how cross-reference indexes are implemented. A symbolic that breaks down the text into it's individual words is then indexed.
Pat
After thinking about it…
This is 'partially' true. The symbolic simply passes the {field} to XREF routine which in turns processes.
Of course .. the simplest example/proof that it does work
Thanks pat
[email protected] onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"
David Tod Sigafoos ~ SigSolutions
Phone: 971-570-2005
OS: Win2k sp2 (5.00.2195)
OI: 4.1
DSig,
When it comes to maintaining indexes on symbolics with values from other tables I haven't practiced myself what I preach. Perhaps in this new app I'm working on.
![]()
- Oystein -
I'm sure Sprezz has covered this ground in both S/ENL and their Knowledge Base. One of the problems is that if your symbolic refers to another table and a change is made in that other table, the symbolic in the original table is not maintained unless you explicitly call it. If this is the case, then it's possible to write your own indexing transaction into the ! file so that it gets processed.
Don Miller
Official RTI Response:
You can put indexing on symbolics.
It works best (most accurately) when the symbolics are self-contained in the file that is being indexed.
You can index on symbolics that go to other tables- Sprezzatura had a RevMedia article on it in 1989 or so- I remember a talk that Andrew gave at a Cogent conference in Florida…
Mike
Thanks .. that will make them very happy ..
[email protected] onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"
David Tod Sigafoos ~ SigSolutions
Phone: 971-570-2005
OS: Win2k sp2 (5.00.2195)
OI: 4.1
Oystein,
Look up Revmedia Vol 2, Issue 4, page 5 -6
This sets out a very simple straight forward method - adding into field 21 of dict (depends on) the following:
FILE*FIELD eg DEPARTMENTS*EMPL_ID
This establishes the dependancy MFS
Mike, Is this sufficiently robust to get doced into Help and or included in any Index build routine?
Richard Bright
BrightIdeas New Zealand
and as it's you Oy… http://www.sprezzatura.com/revmedia/v2i4a3.htm
World Leaders in all things RevSoft
The practice of indexing symbolics is actually punishable by death in the antipodes . We always wanted to do it, but it wasn't sanctioned in the coding practices adoped on most LH-based developments.
As this is not Sprezzatura's official position, nor is it Revelation Software's official position, I will now politely head back to my cave .
I'm certain, DSig, that if you have a need to do it, that it's completely justifiable as an expedient and concise technique.
Steve
World Leaders in all things RevSoft
.. so I will put the ball in your court.
The user has a need to index a field which some 30% of the 10K+ items seem to be Scientific Notation
How would you "seamlessley" make the selections come up? No change to the user interface .. or user procedures.
I, like you, have always stayed away from indexing symbolics although as noted before I think there have been 1 or 2 instances where it became *necessary*.
Of course the best solution would be a "fix" for the situation. Any other ideas are greatly appreciated.
[email protected] onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"
David Tod Sigafoos ~ SigSolutions
Phone: 971-570-2005
OS: Win2k sp2 (5.00.2195)
OI: 4.1.1
PII 300 laptop
Sprezz,
v2i4?
- Oy -
ermmm? Sorry?
World Leaders in all things RevSoft
Sprezz,
Is it really vol 2 issue 4?
(And the link doesn't work.)
- Oystein -
Hey, the link works for me! :-p
World leaders in all things RevSoft
spoken too soon - it works for those of us who got to the page BEFORE we changed ISP and the ISP is now being a little stricter on caps!
http://sprezzatura.com/revmedia/V2I4A3.HTM
World Leaders in all things RevSoft
And for those of us who need to put the www in front
http://www.sprezzatura.com/revmedia/V2I4A3.HTM
World Leaders in all things RevSoft
Sprezzatura,
The reason the link doesn't work here must be that my browser is case sensitive. Your link says . With correct case it should be .
(Also I got completely sidetracked when I saw the "v" and the "i" in "v2i4". I immediately thought of S/ENL.)
- Oystein -
Richard/Sprezzatura,
Thanks. I found and read the article.
The method described in the article uses a relational index. I don't like relational indexes much.
Also it seems to me like a bit of overkill to maintain huge amounts of keys to get this feature implemented. Something that knows about which fields are involved should be enough.
But at least the 64K-of-keys limit on relational indexes must be gone in OI32. In one of the places I (my clients) need this feature there will often be a row or two that is related thousands of times.
Any comments?
- Oystein -
IMHO, I do not like Relational Indexes (takes to long to rebuild, not efficient with large number of keys) and will use them only when there is no other option. If used it should never be used if the list of keys gets into the thousands. This can be better accomplished through the use of a Btree Index with a Btree extract.
Just my two cents.
John
BRAVO!!!
When cross reference was all we had it beat selecting a file for information .. but since btrees there has been no reason to maintain the use of Cross Reference.
Just like indexing symbolics .. you can do it .. but the practice should be avoided if possible.
[email protected] onmouseover=window.status=the new revelation technology .. a refreshing change;return(true)"
David Tod Sigafoos ~ SigSolutions
Phone: 971-570-2005
OS: Win2k sp2 (5.00.2195)
OI: 4.1