Re: Confusing terminology between usher and monotone and proposed change
Richard Levitte <[email protected]>
| Newsgroups | gmane.comp.version-control.monotone.devel |
|---|---|
| Message-ID | <[email protected]> |
In message <[email protected]> on Mon, 9 May 2011 17:09:50 -0400, Hendrik Boom <[email protected]> said: hendrik> On Thu, May 05, 2011 at 10:42:41AM +0200, Richard Levitte wrote: hendrik> > I've had a closer look at the terminology used in usher and in hendrik> > monotone, and there is a part that's quite confusing: hendrik> > hendrik> > In usher terminology, different databases are served by different hendrik> > monotone server, and therefore, the URI to access them through a hendrik> > server name would be expressed as mtn://HOST/SERVER?PATTERN. hendrik> > hendrik> > In monotone terminology, the same URI is expressed as hendrik> > mtn://HOST/PATH?PATTERN. hendrik> > hendrik> > Furthermore, usher is a server in its own right, so when talking about hendrik> > the usher+monotone combination, it might be confusing to talk about a hendrik> > server, as it might not always be clear if you're talking about the hendrik> > usher server itself or one of the underlying monotone servers. hendrik> > hendrik> > Also, in usherctl, the confusion is increased, since it uses PROJECT hendrik> > to designate what usher calls SERVER and monotone calls PATH. This is hendrik> > confusing since monotone has another idea of what a project is, and hendrik> > will just increase as soon as policy branches are in place. hendrik> > hendrik> > hendrik> > To clear the confusion, I propose that we make a terminology change in hendrik> > usher, where the term SERVER (to designate a monotone server entry in hendrik> > the usher configuration) be changed to PATH (with the implicit hendrik> > understanding that a PATH is then served by the monotone server in hendrik> > said entry). hendrik> hendrik> This would presumably be the server name in the "server" line. The hendrik> "foo" in hendrik> server "foo" With my idea (just for the sake of being explicit), the following: server "newpub" local "--confdir" "/home/levitte/usher.projects/newpub" "-d" "/home/levitte/usher.projects/newpub/database.mtn" "--no-standard-rcfiles" "--rcfile" "/home/levitte/usher.projects/newpub/monotonerc" "--timestamps" "--ticker=dot" would be replaced with: path "newpub" local "--confdir" "/home/levitte/usher.projects/newpub" "-d" "/home/levitte/usher.projects/newpub/database.mtn" "--no-standard-rcfiles" "--rcfile" "/home/levitte/usher.projects/newpub/monotonerc" "--timestamps" "--ticker=dot" In human terms, the former would be expressed "this is the server named 'newpub', and local indicates how to start it." The latter would be "the path 'newpub' is served with a monotone started with the arguments given by local." hendrik> But the word "path" is in common usage to a sequence of hendrik> directory names separated by slashes, possibly ending in a hendrik> file name. The word "path" has been expanded, especially if we speak in URI terms, to something of a structured notation to reach a specific resource within a specific realm. That's exactly the way PATH is used in mtn://HOST/PATH?PATTERN . hendrik> This still leaves room for confusion, since (unless I'm hendrik> grossly confused) it's not the file name of the data base hendrik> that's wanted here. No, it's not the file name of the database, but it's a way to reach it. My main issue, though, is that things are expressed differently in the monotone speak and in usher speak, that's where we have a real possibility for confusion. How would you have it? -- Richard Levitte [email protected] http://richard.levitte.org/ "Life is a tremendous celebration - and I'm invited!" -- from a friend's blog, translated from Swedish