Jay, hello. > On 2015 Jan 14, at 13:41, Jay McCarthy <jay.mccar...@gmail.com> wrote: > > 1. Create a directory where the files will go: $ROOT > > I do this anyways, because everything is in git somewhere. > > 2. Make prog.sh in $ROOT that looks like this: > > #!/bin/sh > > cd $(dirname $(grealpath $0)) > racket -t prog.rkt > > 3. Link $ROOT/prog.sh into my ~/bin as 'prog'
Sure -- that's what I was thinking of when saying 'installed a bunch of files in a directory tree along with a launch script' (and using exec, of course). But the problem with that is that it's inherently a rather more ...floppy object. My 'make install' actions would probably do something like install the various .rkt files into ~/local/bin/prog.contents/ and make the link from ~/local/bin/prog -> ~/local/bin/prog.contents/launch-prog.sh. But one doesn't typically expect a script to have a directory attached to it (ie, this smells non-standard, and there isn't really a standard place for such a prog.contents/ directory to go), and it means that if I blow away ~/local/bin/prog I have to remember to delete prog.contents, too. Also if, in this scenario, I delete $ROOT because I think it's redundant and I've forgotten about the link, then ~/local/bin/prog has broken without me realising it. Having ~/local/bin/prog link to the live checked-out files means that I can't hack away at those and be sure that the 'stable' version in ~/local/bin/prog is still working. None of the above are grievous problems -- I could live with all of them -- but they're less neat than 'here is a script file; put it somewhere in your path'. Perhaps the concatenate-with-rewriting is the best solution here, after all. All the best, Norman -- Norman Gray : http://nxg.me.uk SUPA School of Physics and Astronomy, University of Glasgow, UK ____________________ Racket Users list: http://lists.racket-lang.org/users
