On Fri, 11 Sep 2009 19:38:21 +0200, Thomas Berg wrote: >> -----Ursprungligt meddelande----- >> [mailto:[email protected]] För Paul Gilmartin >> >> Is this comparing RECFM=VB to RECFM=FB with equivalent data length? >> I have long believed that this is because ISPF (and XEDIT) >> really don't understand variable length records; rather they >> pad each record with blanks on input to the maximum length; >> proceed as for FB; then strip trailing blanks on SAVE. >> Further evidence: it's surprisingly easy to overflow >> available REGION with a fairly small (on DASD) RECFM=VB file >> with only one very long record and many short records. > >If You with "data length" mean actual, existing data, not the >"LRECL" of the dataset; then Yes. > >I'm using VB 32756/32760 as a way to "catch all" needs. >When I then have 1000 rows in an existing member (this is >a PDSE), it take someting like 3 seconds to repeat a row. >(Line command R.) >Then length of the data in a row is max 80 bytes. > Indeed, I failed to write what I meant, "length in the DCB". Your observation supports my surmise that ISPF (like XEDIT) pads all records to that length.
>This makes it somewhat unusable when You want/need to work >fast. Which I normally do. This ought to be a topic for a performance APAR. You likely would get no better than SUG. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
