On Tue, Oct 21, 2003 at 06:48:39PM -0500, Ben Collins-Sussman wrote:
> Here are some points based on real empirical evidence, rather than
> fuzzy feelings:
> * Subversion has been self-hosting for over two years. Many other
> projects have been using Subversion for nearly as long. What is
> the result of this experiment?
> Answer: global revision numbers work really well; nobody is
> confused, nobody is wishing it were vastly different, everyone is
> used to it. It makes many, many things extremely easy and
> conceptually clear. If the design had backfired, we would have
> gone back to the drawing board early on.
I don't think that's his point. AFAICT, he never suggested that the
global revision number be eliminated.
> * Every once in a while, somebody comes to the list, and says,
> "global revisions are a bad idea", or "I can't live without
> per-file or per-project revision numbers". We tell them to try it
> out, and they come back and say, "hey yeah, it does work."
Or, they don't come back at all. Do you really know that the
conversion rate is?
> * Global revision numbers aren't real revision numbers; they're just
> names for commits/changesets. Instead of labeling them "1, 2, 3.."
> we could have generated names like "blue, red, green". They have
> nothing do with any particular project, so placing N projects in a
> repository is not a problem. When Garrett said it was a matter of
> psychology, he wasn't accusing you of having a psychological
> problem; he was simply saying that new users tend to fixate on the
> revision numbers, giving them more meaning than they really have.
> After a couple weeks of using SVN, the attitude changes, and users
> come to understand revisions as mere commit-labels. That's the
> "psychological change" we're talking about.
> There's a certain absurdity to the situation here: I'm seeing one
> person who has never used Subversion, trying to argue to dozens of
> satisfied users that the model is deficient in various ways.
I'm with you on this one.
> I'm not sure what to tell you, except, 1) read the book, and 2) try
Is is really a bad idea to be able to put a keyword entry in a file
that is a file-relative change count?
To unsubscribe, e-mail: email@example.com
For additional commands, e-mail: firstname.lastname@example.org
Received on Wed Oct 22 02:00:03 2003