Branko Čibej <br...@wandisco.com> writes: >> We could declare want-iprops to be an error and remove it, that also >> risks breaking third party client. We could change the documentation >> of the protocol to match the released implementation, but that would >> mean accepting the "? foo" form rather than "(? foo)" and I seem to >> recall that the later is preferred as it's easier to extend. > > Unless the documentation is clearly bogus (and in this case, it's not), > our implementation should match the documentation. The only problem is > that older clients might not work with a 1.9 server; but there's not > much we can do about that, short of accepting both forms of the command, > and I think that would be worse.
I think the documentation might be bogus. The released 1.8 server parses get-dir using the format: "c(?r)bb?l?B" where "?B" is optional. Implementing the documented format would be: "c(?r)bb?l(?B)" but I think that would break old clients that send "l" but do not send "(?B)", i.e. our clients, as well as breaking third party clients that try to use the released format. I suppose we could attempt to "correct" both the documentation and the implementation to use the format: "c(?r)bb?l?(?B)" which would work with our old clients, but that would still break any third party clients trying to use the released 1.8 format. -- Philip Martin | Subversion Committer WANdisco // *Non-Stop Data*
