Re: [Issue 1039] Changed - ra_dav / svn cat : adds a blank line at start of file

[Greg Hudson <gh...@MIT.EDU>]

On Sun, 2002-12-22 at 01:56, Peter Davis wrote:
> I can't help but think that, barring line-buffer issues, the newline could 
> just be echoed to stderr instead of stdout (same for the password prompt).  
> After all, that is the purpose of stderr: a stream for messages that 
> shouldn't be redirected when redirecting the output into a file or pipe.  Has 
> this already been discussed?

It's also common to use /dev/tty for password prompts, in case input has
been redirected (ssh does it, for instance, which is why you can use ssh
with password authentication as an ra_svn tunnel agent).  Perhaps we
should do that, and send the echoed newline there as well.

(That's Unix-specific, of course.)


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: [Issue 1039] Changed - ra_dav / svn cat : adds a blank line at start of file

[Karl Fogel <kf...@newton.ch.collab.net>]

Greg Hudson <gh...@MIT.EDU> writes:
> It's also common to use /dev/tty for password prompts, in case input has
> been redirected (ssh does it, for instance, which is why you can use ssh
> with password authentication as an ra_svn tunnel agent).  Perhaps we
> should do that, and send the echoed newline there as well.
> 
> (That's Unix-specific, of course.)

Turns out apr_getpass() already does this (my comments in the issue
were wrong).  It looks like we introduced the spurious newline in very
old commits in subversion/clients/cmdline/trace-commit.c (revisions
1.7 and 1.8).

That's from CVS days, and trace-commit.c is gone now, but maybe that
newline lives on somewhere.  Diagnosing now...

-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org