This is a question on best practices and prudent programming.
I'm importing text data from a Big Company for updating a client's item
table.
It's a fixed field text file and the specs for the data file specify max
line length of 700 characters.
The test file I'm using is 15K lines (8.5 Mb).
I import the data into a temp table:
CREATE TEMP TABLE `RawPkgSpecTmp` +
(`RowNbr` INTEGER, +
`PkgSpecData` TEXT (700), +
`ColLength` = (SLEN(PkgSpecData)) INTEGER)
AUTONUM RowNbr IN RawPkgSpecTmp USING 1
--Load data
LOAD RawPkgSpecTmp FROM PkgSpec.TXT +
AS FORMATTED USING PkgSpecData 1 700
I cursor through the RawPkgSpecTmp table a row at a time for processing,
loading the column PkgSpecData into the variable vPkgSpecData, then
using SGETs to extract relevant data.
Most of the rows are 446 characters, I assume there's a CR at the end of
the data, not position 700.
Occasionally the data length is 668 characters and there is a clump of
data at position 661 that I need.
So the question is:
Is it safe to always do a SET VAR vXtraPartNbr =
(SGET(.vPkgSpecData,8,661)) even if there are only 446 characters in
vPkgSpecData?
Or, for safety, should I check the length of the row first:
IF vColLength = 668 THEN
SET VAR vXtraPartNbr = (SGET(.vPkgSpecData,8,661))
ENDI
to make sure I'm not wandering off into the data-hinterlands that could
back to bite me someday?
R:Base doesn't seem to mind - vXtraPartNbr is an expected null when the
data is short and there aren't any apparent issues, messages or error
messages.
Just asking, FMI.
Doug
p.s. This question came about because I was getting "Out of Dynamic
Space" errors. The error was related to a subsequent command and
totally unrelated to the SGET. But it got me wondering...
