On Wed, 18 Feb 2009 20:11:53 +0000, Ted MacNEIL wrote: > >>Also there has been some extra protection for a *long* time that keeps two >>tasks from opening a PDS for output (IEC143I 213-30 abend). > >When you edit a PDS member, the file is not open for output. The member is >just a memory copy. >It's opened for output, when you save it. >ISPF keeps information in memory, but does not re-read the directory. >So, if two people (or more) are editing members, directory already read, the >last saver 'wins'. > Ummm. If two users edit the same member of a Classic PDS (not PDSE), not concurrently but consecutively, with no overlap, it's got to re-read the directory (or at least do a BLDL) because the STOW associated with the SAVE changes the TTR. Otherwise the design is very wrong. With a PDSE, if the second user has done a BLDL, the directory change will not appear until he does a CLOSE. Then the BLDL result becomes invalid and another BLDL is needed.
>Try it! You won't like it. > >I did a test with some of our developers, just about two years ago. > >The data was corrupted. > If by corrupted, you mean "last person wins", that's WAD. When I update a shared PDS, even in batch, I use ISPF LM services. I've experienced one instance of corruption, when MIM had crashed on the system the other lady was using (ISPF EDIT of a different member; I found her member with my member name; identified her from the member content and asked our sysprog what might have failed.) -- 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
