RE: Creating parallel pipes
It is called an "affixed pipeline". Add unconnected streams to the
stage when it initialises, creat the pipeline with ADDPIPE and connect
to the added streams.
Then do OUTPUT to emit a record into the affixed pipeline and read the
response with READTO. The challenge is to either ensure that you
always get a one-record response (or you always know how many records
to read).
Take a LOOKUP stage as an example.
'addstream both'
'addpipe (end \)',
'\*.output.1:',
'|copy', /* Release producer! */
'|l: lookup w1 master',
'|insert /0/',
'|i:faninany',
'|*.input.1:',
'\<....',
'|l:',
'|insert/1/',
'|i:'
Then when you want to look up something in your REXX code, you can
'select both 1'
'output' query
'readto resp'
parse var resp found+1 data
I've not been able to measure any improvement over repeated reading of
a file from loading it into INSTORE and then getting the contents from
OUTSTORE. ymmv.
j. [EMAIL PROTECTED]
