Thanks for the input guys. I must confess, I "knew" the answer, but after RD's additional info, I didn't know it was "that big" of a hit. However, for this one-time import deal, I still chose to use the GU route for my own peace of mind to not lose the plot with any @COMP or fieldname issues. Customer reviewed the rules, and elected to cut down on a date range thereby excluding several million reads (i.e. 0.5 mill x "lots") - runtime is about 10-minutes. I may rewrite it all as an LU one day, for comparison purposes.
@kapoof/Rob - I haven't as yet played with Child filedefs (though was aware of their existence), and given time constraints, didn't feel it was a good time to learn about them - they do seem useful though.
@RD - I have considered readonly bootstraps - and in the perfect world of never Production coding or hot-fixing it'd be great - but...can't say never at this client's site
Regarding alternate files, what version are you on that alt files cannot be automatically created?
6.2. I guess it depends on what "automatically created" actually means...for this one-time deal, I have created a few alt filedefs for some of the files, and a quick'n'dirty utility to rename the field names in filedef and vipbm25. I'm basically - perhaps somewhat oxymoronically - doing "extra work" in the name of safety but being a lazy b*stard don't want to do too much extra work...
Under some time pressure (i.e. by about next Tuesday) to get this done (including testing) - but at the same time, going for a little education I'm pretty much done now - onto some testing soon.
Here's the guts of the scenario: - client is pure pro-isam - have 1 progdir and multiple datadirs, each with a different @COMP - based on some user-defined simple and complex rules, have to pull some data from each of the datadirs, and drop them into yet another datadir with yet another @COMP - I have a personal dislike of screwing with @COMP (call it paranoia) - data is always going from FILE-A in datadir1,2,3 to FILE-A in new datadir - hundreds of fields involved - some being directly imported, some "if rules" based on source datadir, some arbitrary hardcodes, and some "auto numbering"
I do have to say, I doubt there would be any way possible for me to get this done in the timeframe involved, if I weren't working in ProIV - perhaps even if I were working with an RDBMS. $0.01
Thanks for your reply. It's only 0.5 million calls each making a totally unknown number of reads. This is envisaged to be a "one time" deal, so maybe "cost" doesn't matter quite so much, upon reflection vs the human cost (mine) of creating a not insignificant number of extra alt-filedefs.
Have you - or has anyone - done any "clinical math" on this general scenario? I'm thinking maybe I should one day just to get a better feel for it.
I have read a little on that function caching in the doccos - but am going to wait until my client gets a secondary server (lesser spec than primary) up and running before I think about experimenting with it.