Dan Hain wrote: > Sebastien Roy wrote: >> seb-1: >> >> "Unlike ping(8) , fping is meant to be used in scripts and its >> output is >> easy to parse." >> >> If it's meant to be used in scripts, then why is the output format >> "Not an Interface"? > Ok, it's an Interface. Should I include sample output from the various > option mixes? I'm not sure how to document the interface.
At a minimum, the stability level of the output format should be specified here. If you can also give sample output here, then that's great, but it's more important that the actual format that script writers can depend on be documented in the end-user documentation (i.e., the man page). >> seb-2: >> >> "-d Use DNS to lookup address of return ping packet." >> >> Solaris has a name-service switch which allows the administrator to >> configure a naming service on a system-wide basis, which commonly-used >> APIs like gethostby*() and get*info() functions adhere to. Does fping >> really bypass all of that and use DNS directly? > No, both the "-d" and "-n" options use gethostbyaddr() So the man page is outright wrong. Please fix that. >> seb-3: >> >> The spec states: >> >> "The program uses a raw socket to communicate and requires root >> access >> (geteuid() == 0)." >> >> Yet the man page states: >> >> "-f Read list of system from a file. This option can only be >> used by >> the root user. Regular users should pipe in the file via >> stdin: >> fping < targets_file" >> >> That implies that users other than root can run the command. Which is >> accurate? > > The effective uid must be root to run the program. In addition, if the > "-f" flag is specified, you really have to be root. This issue seems to have spawned questions from others, so I'll let you drive the issue to completion in those sub-threads. I have similar concerns to those mentioned by others. Thanks, -Seb
