On 2016-02-15, at 01:32, John P. Hartmann wrote: > The name can be a word of any length, but as the message buffer is 256 > bytes, you'll never see more than about 230 of them. > Wonderful! I can incorporate a time( L ) in the name and get (almost) unique names. I don't think the code path through ADDPIPE can take less than a microsecond. Unless Pipelines exploits multiple processors for concurrent stages. I suspect it doesn't.
On 2016-02-15, at 01:02, Rob van der Heij wrote: >> Thanks. I see why I should have been maming all my pipelines. (BTW, >> what's the syntax of a "name"? I don't easily find it in the manuals. >> Can I use 100 characters?) >> >> Is there a command that will list all active stages? > > Your book must be missing pages. Chapter 21 section "Options" in mine has > the globalOption "NAME" that takes a word as the argument. The only entry in > the index for NAME points to there too. > It can be real hard to find a publication number nowadays. Looking at: z/VM 6.3.0>Operation and Use>z/VM: CMS Pipelines Reference>Pipeline Subcommands>ADDPIPE https://www-01.ibm.com/support/knowledgecenter/SSB27U_6.3.0/com.ibm.zvm.v630.dmsc3/leaddpip.htm I see essentially: ( NAME name ) Where the descriptive text provides effectively a circular definition of "name". If I could find a publication number, I'd submit an RCF. > If you would want to use "jeremy" to show the pipeline set (your 2nd question) > you should put distinguishing aspects in the first 25 positions of the name. > > PS When I forget the SEVER in the program, I can reproduce the problem. Doing > a CMS STORMAP every few iterations shows that each file open takes 64K under > the line, and that's what runs out in my case. But with SEVER to wrap up the > pipeline and close the file, the problem is gone. > It was a novice mistake, found more by inspection than by tracing. I copied two lines of code and failed to change the stream name (but I did change it in the comment). Thanks, gil
