🚀 go-pugleaf

In article <9201180109.AA05273@ucbvax.Berkeley.EDU> you write:
>
>1) Reads specific records from 1,000,000 record file(s).  I would pass
>   the line record number(s) that I wanted read as parameters to the
>   program.  It then puts these records into the console stack.  (I know
>   that this is basically an easy program to write, but I am busy with
>   writing a larger application that needs two specific assembler files,
>   and I would really like a 'gift' from the network instead of having
>   to write my own from scratch.)
It will be only slightly slower to use the basic facilities
which CMS already provides, e.g., VM/SP System Product Interpreter
and native CMS commands, i.e.,
   /* REXX, of course */
   /* Usage: 'READIT' record numbers are placed here,
      e.g., 'READIT 1 33 555 7777 99999'
      Resulting records are placed into the stack */
    arg parameters
    file = 'HUGE FILE A'
    address command
    do while (parameters <> '')
      parse var parameters token parameters
     'EXECIO 1 DISKR' file token
      if rc = 0 then iterate
      say 'Unexpected return code' rc 'when attempting to read record' token
      leave
    end
    t=rc
   'FINIS' file
    exit t
>
>2) Write a three byte binary value to a file.  It increments this value
>   by one and then writes it to the file (along with some other data).
>   The record layout looks like "AAABBBBBBBB", where "AAA" is the
>   binary field, and "BBBBBBB" is the other field.  "AAA" starts at '1'
>   and is incremented up to about 1,000,000.
/*  REXX, again */
 address command
 arg integer other_field
 RECORD = d2c(integer+1,3) || other_field
 file = 'TEST FILE A'
'EXECIO 1 DISKW' file '( VAR RECORD'
 exit rc

Note that this code writes the new record to the "end" of the CMS file,
and that, for efficiency sake, does ***NOT*** issue "'FINIS' file".
This is left as a responsibility of the problem which calls this EXEC file,
i.e., to take a check-point every "N" records.

>             (think of doing an XEDIT search on "/DOCTORS" in a 1,000,000
>record 200 byte CMS file versus the same search on a 15 byte file)
>
Actually, XEDIT uses the translate-and-test (TRT) instruction to search
each record for the first character of the search-string, e.g., 'D'.
Thus, one machine-language instruction will very quickly eliminate
the need to intensively-search many of the record, or it will give
a suitable column-pointer for the start of an intensive search.