On Tue, Feb 14, 2006 at 12:54:50AM +0100, Marc 'HE' Brockschmidt wrote:
> > Oops, sorry. The things which are not related or shouldn't be there:
> > - the "my $caller = (caller)[2];" line in get_ssh_connection; that's
> > cruft from my debugging
> > - the change of $& to $1 in line 1460:
> > $remote_ssh_port =~ s/^\s*(\d+)\s*$/-p $&/;
>
> Yeah, plus a protocol addition which seems quite unnecessary to me.
OK; see below.
> > An altogether different approach would be to log the communication at
> > the local end rather than at the remote end.
>
> Actually, the only reason for having this code in the released version
> is that I was too lazy to add/remove it when developing. I used it when
> somehow the "server" didn't behave as expected, but as I don't expect
> large changes to the codebase in the near future, I'm more inclined to
> remove the whole log feature than loosing compatibility to the version
> released in sarge.
Either remove it or just comment it out, or just remove the --debug
option from the getopt invocation. I agree, that would be much more
sensible.
> >>> Other comments while I'm here:
> >>> (1) You don't need the & in front of sub names; they'll work quite
> >>> happily without it.
> >> I know that quite well, but this is a matter of personal coding style.
> > Indeed. The only significant difference is that if you want to use
> > prototype checking, you cannot use the &.
>
> I looked this up, BTW, and the reason for the & use is that it is used
> for the "send" and "read" functions, which are also defined by
> perl. Perl uses the global definition without the & in front of the
> name.
Ah, hadn't spotted that one!
> >>> (2) -o for a remote ssh port number is a horrid choice: -o is normally
> >>> an output file; -P would be much better (matching scp, for example).
> >> -P was my first choice, but due to the limitations of the used
> >> Getopt::Long module (which is not able to see the difference between -p
> >> and -P) and the already existing "-p" short option for
> >> --cache-passphrase, something else needed to be chosen. As all other
> >> choices are equally bad, I used -o.
> > Getopt::Long::Configure("no_ignore_case");
>
> I remember trying this option, but something was weird [tm]. I'd need to
> look this up, as I'm sure I ranted in IRC about it, but there were
> definitely some problems.
Weird indeed. It seems to work for me.
Julian
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
