> 1) Write small compatibility shell scripts that invoke 'hostname' with > the right parameters. For example, the 'dnsdomainname' script would > invoke 'hostname -d'. > > This would cause problems if you do --disable-hostname > --enable-dnsdomainname; and where the system hostname does not provide > the -d flag for whatever reason or prints the results in a non IU > fashion.
Yeah, that approach would require that if any of the *name scripts are installed, the hostname binary needs to be installed. > 2) Extend the 'hostname' tool to allow that argv[0] influence its mode > of operation. This is how Net-Tools work today, they have only one > program but it is hardlinked to all these names. Letting argv[0] > influence mode of operation is a bit ugly, but there are some precedent > for it even in well-maintained projects like CoreUtils. > > I'm quite sure coreutils doesn't depend on the value of argv[0]; it is > also discouraged in the GCS. At least the 'test' tool inspects argv0 to see whether it should operate in normal test mode or the [ mode. > 3) Install separate tools that implement the operation. This would > result in better documentation, better --help output, etc. Technically > the tools could share most of the code and simply use some #define and > #include's for the gist of the operations. > > This is my favourite one; this would be very similar to true/false in > coreutils. But requires the most work. It is also my preference. Unless there are other voices within a few days, I'll propose something along these lines. Additional thoughts or comments still welcome! /Simon
