> I noticed some problems with the way the command line
> client was checking the user's input before passing
> commands to the actual function. For example, a
> call to `svn delete` should notice that at least
> 1 argument is required and print the help text
> for the delete subcommand.
Well, we could always just have it print out the command specific help...
> The current code passed the args to the delete subcommand anyway and
> that printed the main usage text instead of the delete help
> text. The following patch fixes the problem by adding a min and max
> number of targets field to the command struct. Main then checks to
> see if the number of arguments the user passed satisfies the (0 to
> N) or (1 to N) requirement.
Is this really worth it? Looking through the commands that we need to
support, it seems like only a few have this requirement:
Where 'Number of file args' is blank, 0, 1, or more files can be
passed. This includes commands that take '.' as a default file arg.
COMMAND Number of file args
add 1 or more
copy 2 or more?
delete 1, or more
log 0, 1, or more
I would vote to leave the checking in the command, although I don't
feel all that strongly about it. Does anyone else have any feelings on
Received on Sat Oct 21 14:36:25 2006