Re: [PATCH] Don't force imap.host to be set when imap.tunnel is set
- From
Jeff King <peff@peff.net>
- Date
- Apr 22, 2008, 10:41 UTC
- Message-ID
- <20080422104146.GA11198@sigill.intra.peff.net>
- In-Reply-To
- <fuka50$pce$1@ger.gmane.org>
On Tue, Apr 22, 2008 at 10:11:59AM +0100, Andy Parkins wrote:
Show 5 quoted lines
> I'm sure you are correct, but as I say - it's not guaranteed. Since > git-imap-send can't know what this particular tunnel requires it shouldn't > force the creation of a dummy option. If the tunnel does require a > hostname then there is a place to put it, and the person writing the tunnel > line can decide that.
I think Junio's point is that it's easy to start dereferencing NULL, because later parts of the code assume that "host" is always set, even if only to use it for informational purposes. So those callsites either need to be fixed to handle a NULL host, or perhaps something like this instead (totally untested):
diff --git a/imap-send.c b/imap-send.c index 04afbc4..db65597 100644 --- a/imap-send.c +++ b/imap-send.c @@ -1303,8 +1303,11 @@ main(int argc, char **argv) return 1; } if (!server.host) { - fprintf( stderr, "no imap host specified\n" ); - return 1; + if (!server.tunnel) { + fprintf( stderr, "no imap host specified\n" ); + return 1; + } + server.host = "tunnel"; } /* read the messages */ -Peff