Re: [RFC] Make svnserve require --inetd to use stdin/stdout

On Sat, 2004-01-03 at 11:30, Ben Collins-Sussman wrote:
> I dunno what Greg Hudson thinks, but is this really a problem? Have you
> seen all sorts of people run 'svnserve', see spew on stdout, and then
> get confused, wondering if it's running or not? I certainly haven't.

I guess it's a slight problem. I've seen some people configure sshd so
that the command becomes "svnserve" instead of "svnserve -t". This
situation works too well; instead of a usage message, you get a valid
connection, only you're not authenticated.

I've also certainly seen people run svnserve and be confused by the
protocol spew. A usage message might be enough to get them going, even
if they'd be better off reading the book or the man page.

My main criticisms of Tobias's patch are:

  * The main thrust of the patch belongs in 1.0 or not at all; it is not
okay for 1.1 to break people's svnserve-in-inetd setups. So it should
not be committed to the mainline right now; it should be filed as an
issue and entered as a 1.0 candidate.

  * The --foreground option does not deserve a single-letter option
name. (Neither does -X at this point, although it has a parallel in
httpd. And probably not -T either. But those are there historically
and it's a pain to get rid of them.)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Jan 3 17:50:00 2004