On Fri, 15 May 2020, at 19:48, Charles Mills wrote:
> But everything to do with how big a stem variable array it can build.
But there's no need, if counting the records in a file (or even summing their
lengths as well) to store them all. Use something like:
/* REXX
Call as: %linkount dsn num
where: dsn is fully qualified, with/without quotes
num is number of lines read at a time, default is 500
*/
parse arg dsname chunksize ; dsname = strip(dsname,"B","'")
if ^datatype(chunksize,"W") then chunksize = 500
address tso "allocate fi(huge) da('"dsname"') shr reuse"
linecount = 0
do forever
"execio" chunksize "diskr huge (stem somelines."; erc = rc
linecount = linecount + somelines.0
if erc > 0 then leave
end
"execio 0 diskr huge (finis"; "free fi(huge)" /* close+free */
say dsname "has" linecount "lines."
One could possibly preface that code with one of the LISTDSI-based
solutions to get an approximate idea of the allocated size of the file
& hence determine an appropriate chunksize, depending on whether
you want the count to be faster/slower or to waste more/less memory
while running.
--
Jeremy Nicoll - my opinions are my own.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
