Heads up on "reindex"....
Jay R. Ashworth
jra at baylink.com
Thu May 10 07:22:20 PDT 2007
On Thu, May 03, 2007 at 06:48:18PM -0400, John Esak wrote:
> My feeling is that if you need to rebuild all the indexes, say you're
> copying a file to a new name or some such, you can work off a list/table for
> original index values, but what if the users have *added* brand new indexes
> of their own? For the new SS Accounting, I wrote a very elaborate (read
> thorough) menu-driven indexing routine which deals with files, qualifiers,
> etc. Whatsever in the list gets done, whatever it finds new gets done
> also... all with file i/o calls to read the actual index headers directly so
> it would work on Windows and *nix. It also copies key/data if required.
That's extremely cool... but one consideration seems important, and I
haven't looked closely enough to know whether you've addressed it or
not.
While filePro makes it easy for end-users to *add* indexes to files to
make it easier for them to search for things 'their way'... it's always
annoyed me (WISHLIST) that there is no way for application writers to
*lock* indexes so that the end user cannot touch them -- since as we
all know, extending an index *one character* can cause an entire
application to blow sky high.
If your index manager doesn't already have a facility for this sort of
locking, you might want to think about adding one. Additionally,
FWIW, we've had situations where both manual index rebuilds and -ra
rebuilds failed to clear certain types of index-related problems... but
*whacking* the index and rebuilding it from scratch worked.
I have never built a reasonable mental model of why that might be
possible, but by inspection, it sometimes happens. Another issue to
conside in reindexing programs...
Cheers,
-- jra
--
Jay R. Ashworth jra at baylink.com
Designer Baylink RFC 2100
Ashworth & Associates The Things I Think '87 e24
St Petersburg FL USA http://baylink.pitas.com +1 727 647 1274
More information about the Filepro-list
mailing list