In a recent note, Art Celestini said: > Date: Mon, 31 Jul 2006 10:52:03 -0400 > > From what I am familiar with from the Linux world, the method does not > appear to be 100% bulletproof. Essentially, the "diff" program (e.g. > http://www.die.net/doc/linux/man/man1/diff.1.html) is used to create > a "patch" file by comparing the old and new source. Then the "patch" > program (e.g. http://www.die.net/doc/linux/man/man1/patch.1.html) can > be used to apply the changes to another copy of the old source. > You needn't search so far afield. Both are part of the base z/OS, likely since MVS 5.2.2. But the GNU/Linux flavors have extensions which I value, and which the open source community regularly exploits.
> The person who runs the "diff" program must ensure that the "context" > parameter specifies enough lines to ensure there are no ambiguities. > Problem is, there is no guarantee that ambiguities won't still exist > in different versions of the old source. I've seen claims that a > particular patch will apply to multiple versions of a target program, > but usually, the patch is documented as being applicable only to a > specific version. > And neither method protects against overlapping patches by different developers. In fact, if two different developers insert similar new code but the sequence numbers by happenstance are different (perhaps differentiated by ISPF's use of 79-80 as a vesion indicator), won't IEBUPDTE cheerfully interleave the patches with nary a warning? patch-diff's verification of context will likely result in at least a warning. -- gil -- StorageTek INFORMATION made POWERFUL ---------------------------------------------------------------------- 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
