[David James]
> Quoting my original mail:
> >> The following modes are supported for directories:
> >>
> >> "+...": All children of this directory are included
> >> by default.
> >>
> >> "-...": All children of this directory are excluded
> >> by default.
> >>
> >> "+*": Immediate children of this directory are included
> >> by default. Subdirectories are included by default.
> >>
> >> "-*": Immediate children of this directory are included
> >> by default. Subdirectories are excluded by default.
>
> We should also add a fifth mode for directories:
> "+.": This directory is included by default, but its children
> are not.
I don't understand the difference between "+..." and "+*". Both of
them include a directory's children - files and subdirectories.
Also, that "-*" functions as a way to include things is unintuitive.
Also, I think "+..." should be written as "", which is 4 characters
shorter.
> If the user types svn viewspec --remove somedir/*, and their shell
> has expanded the asterik, we should display an error message which
> notes that the glob was expanded, and only proceed if the user
> supplies the force option. (We can detect that the user supplied a
> '*' glob by noting that *every* non-hidden file and directory was
> supplied on the command line.)
Ugh. That is way too magical, and it's a bit insulting for a program
to presume to tell me that because it's possible that I used a shell
glob, I probably didn't know what I was doing. What if my shell glob
was *.php and I'm running the same command in several places, and this
particular directory has only *.php files in it?
Surely it is outside the scope of the svn client to lecture users on
how shell globbing works. Can't we trust users to know how to use
their own shells?!?
- application/pgp-signature attachment: stored
Received on Mon Jul 17 00:19:24 2006