Stephen Hahn wrote: [CC:'ing shell-discuss at opensolaris.org to get some feedback from there] > Please read > > http://blogs.sun.com/sch/entry/verexec_1_a_simple_execute > > for some background, and then review > > http://cr.opensolaris.org/~sch/on-verexec/ > > I'm debating implementing some of the refinements mentioned in the > blog entry, as well as providing manual pages for both verexec(1) and > isaexec(1).
1. What do you do if a script clears it's environment, e.g. removes all environment variables except those it thinks are "safe" ? 2. How wide will this be used, e.g. which utilties/commands do you target with this ? 3. I have two concerns about performance: - The extra |exec()| will be a huge pain for larger machines, e.g. on a fully-loaded SF15 an |exec()|-storm can bring-down a machine to it's knees because each |exec()| will do crosscalls to all other CPUs to tear-down the address space and do other housekeepingt stuff. - The extra time spend in the "versexec" wrapper may cause catrastrophic performance problems if this is used for utilities which are expected to run _fast_, e.g. "mkdir"&co. Remember we already have that problem with ksh93's use of "isaexec" and the _only_ way we got away with that was that ksh93 is a big shell interpreter with long startup time (making |exec()| only a fraction of the total startup time) and that the use of even a single builtin command compensates for the isaexec overhead. But if this is used for short-lived things the performance penalty may be a huge price to pay... ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) roland.mainz at nrubsig.org \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 3992797 (;O/ \/ \O;)
