> It does, of course, lead to the question of why you want to know!
I used to have a program I called that would take a select list &
sort it according to the group order of the file it was going to be
applied to. For really big lists it made a really big difference
because subsequent file i/o flowed sequentially through the file with
several consecutive groups read into memory by the os, (yes, except for
overflow) then all records from those groups processed.
I would use it whenever I had a big select list that was built
independently of the file it was going to process and the processing
order did not really matter.
It worked on Pick, where their one algorithm is well-know and easily
duplicated, and on Prime, which gave us the !HASHID function. (I think
that's what it was called, but I may be remembering !HUSHIT, which my
dyslexic brain would always turn around, sticking the "!" on the back
end and associating the "SH" with the "IT" that followed instead of with
the preceding "HU", which dyslexia somehow converted into "OH", anyway.)
I wouldn't care about their algorithms, per se, but I wish IBM would
give us a similar function:
GROUP.NO = HASHID( TYPE, MOD, SEP, ID )
Sort of like the RECORD verb, except that actually goes out and does
>RECORD VOC ASDASDF
Record "ASDASDF" hashes to group 233 but was not found.
>RECORD VOC SYS.HELP
Record "SYS.HELP" hashes to group 847 and was found.
u2-users mailing list