Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!uunet!mcsun!ukc!icdoc!dbh From: dbh@doc.ic.ac.uk (Denis Howe) Newsgroups: comp.sys.acorn Subject: Re: Pseudo Filing Systems via *set Message-ID: Date: 6 May 91 11:52:51 GMT References: <3089@krafla.rhi.hi.is> <1991May1.184943.1835@ugle.unit.no> <3106@krafla.rhi.hi.is> Organization: Dept. of Computing, Imperial College, London, UK Lines: 30 In-Reply-To: thsa@rhi.hi.is's message of 4 May 91 15:11:11 GMT In article <3106@krafla.rhi.hi.is> thsa@rhi.hi.is (Thorvaldur S Arnarson) writes: >This opens up the very lively discussion of globing. Why shouldn't the >cli do globing a la unix shells. Comments anyone? I was disappointed at first that RiscOS's CLI doesn't glob like Unix but there are a couple of advantages to leaving it up to the application that I can think of (gosh, me criticise Unix, This RiscOs must really be getting to me!): 1. The application can tell that it has wild cards in the argument. This makes possible things like mv *.o *.obj which are much harder under Unix. It can also complain if it doesn't want wild args, like *Delete. Of course you can quote the wild cards so that they are passed unexpanded but then Unix doesn't provide you with a nice library routine to expand them (or does it? apart from running a shell to do it :-(). 2. Neither the CLI nor the application need to allocate space for the whole expansion. Under RiscOS you can read one match at a time. Unix has a (large) limit on the length of the argument list. (The Unix utility xargs is a partial solution). I have always suspected the reason this, and various other things are the way they are is that the authors of RiscOS came from a VMS background rather than a Unix one. :-) Am I right? -- Denis Howe C-C H558A Imperial College London C-C-C +44 (71) 589 5111 x5064 N=N