On Thu, Feb 17, 2011 at 5:04 AM, Stephan Beal <[email protected]> wrote:
> On Wed, Feb 16, 2011 at 6:25 PM, Richard Hipp <[email protected]> wrote: > >> My current approach is to run ssh to a shell on the remote side and then >> issue shell command over the wire that invoke Fossil multiple times to act >> as a server for the client-side protocol. It's all a bit dodgy. >> > > git uses this approach as well, as far as i can determine, and i have found > that this approach (which seems sane to me) breaks when hosters (like mine) > hard-code the PATH to /usr/bin:/bin for non-interactive shells. That of > course breaks when fossil/git/whatever is installed in ~/bin (as on my > setup). Because of fossil's built-in CGI support, it's still the only SCM > i've ever been able to get running on my hoster (where i cannot run custom > server processes). > If you fossil binary is not on PATH, you can tell SQLite where it is as a query parameter on the SSH url. For example: fossil sync ssh://userid@hoster/repos/x.fossil?fossil=~/bin/fossil > > -- > ----- stephan beal > http://wanderinghorse.net/home/stephan/ > > _______________________________________________ > fossil-users mailing list > [email protected] > http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users > > -- D. Richard Hipp [email protected]
_______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
