On Fri, Oct 19, 2001 at 01:55:12PM -0700, Greg Stein wrote:
> On Fri, Oct 19, 2001 at 03:46:05PM -0500, email@example.com wrote:
> > Greg Stein <firstname.lastname@example.org> writes:
> > > I second Brane's sarcasm here. We simply will *not* support files which
> > > contain newlines in them.
> > >
> > > Thus, the one-per-line is the best option.
> > Um, programmers (Makefile writers) aren't the only people who use
> > version control systems.
> And non-programmers are the *last* people to put a newline into their
> filenames. I can't even begin to think how a non-programmer would do such a
touch "$(echo -e '\n')"
> Have you *ever* seen such a file? I sure haven't seen one in the twenty some
> years I've been looking at filenames. Not even an accidental one. I did just
> verify that it is possible (had to go into the Python cmdline to do it), but
> seen one out in the wild? Nope.
One easy way to get them by accident is to, in perl for example, read a file
with a list of file names to be created, but forget to chomp() before creating
I have seen lots, but only accidental ones, and those created by people
trying to exploit vulnerabilities in certain cron jobs. They're just too
much of a hassle to actually *want* one.
> > There are currently problems in Subversion which make versioning files
> > with newlines in their names difficult, but that's not the same as a
> > policy of not supporting them. We've not had such a policy before
> > now, as far as I can remember, and (Greg's assertion notwithstanding)
> > won't make such a policy without at least some discussion.
> Set your policy however :-) ... I'm just voting. And mine is a big +1 on not
> supporting newlines in filenames.
> It isn't possible to veto a concept -- only code. But if the code to support
> newlines is hairy... well :-)
I agree completely. Personally, I'd be happy with either of the two patches
I submitted, both of which support whitespace (and newlines) in file names
via use of wildcards. :-)
> > They can always be escaped ("\n") in things like status output,
I like the way the filesystem deals with it: create a file with a newline
in it and suffer the consequences.
> Do you want to do all the work and maintain the stuff into the future? And
> for what benefit? So we can *theoretically* support a case which nobody will
> *actually* use?
Right, I think that work would be misplaced effort.
To unsubscribe, e-mail: email@example.com
For additional commands, e-mail: firstname.lastname@example.org
Received on Sat Oct 21 14:36:45 2006