On Mon, 8 Apr 2013 05:49:58 -0700, Phil Smith wrote: > >In addition, I suspect/believe that update-by-replacement encourages more >changes than necessary: that is, for 30 years, when I've made a change using >XEDIT in UPDATE mode, I've ALWAYS looked at the resulting update before >committing it, to make sure that I hadn't inadvertently/sloppily changed >anything else (e.g., accidentally changed a comment, where "changed" might >mean "typed over it and put it back, but XEDIT saw it as a change"). > I consider this a bug in XEDIT. Accordingly, when I've made CMS UPDATE files, I've always ignored XEDIT's output and used SuperC to generate the UPDATE files.
>Add in the inevitable tendency of diff to pick the wrong lines as being >changed (blank line/close curly braces, for example-this isn't diff's fault, >it's that it can't tell) and diffs aren't even as useful as updates. > The Wikipedia article I cited earlier mentioned this as a hazard of diff's "longest common subsequence" technique. SuperC's "longest common substring" might be more resistant to this, yet more susceptible to other undesired matching. >Since I started working on z/OS stuff, I've been disappointed that there's no >equivalent of XEDIT in UDPATE mode, and that IEBUPDTE isn't compatible with >CMS UPDATE. I find that amazing-the productivity of being able to Just Edit >The Damned File is huge. Maybe IBM has an internal editor they use on z/OS >that creates IEBUPDTE-format updates a la XEDIT? > And worse yet, SuperC with UPDMVS8 option demands that both OLDDD and NEWDD have valid ascending sequence numbers. It's common for OLDDD but not NEWDD to have valid sequence numbers, which is acceptable with the UPDCMS8 option. So to create IEBUPDTE input, I run SuperC with UPDCMS8 and use a Rexx filter to convert its DELTA to IEBUPDTE format. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
