On Sun, Jan 10, 2010 at 02:31:33PM +0100, Roland Clobus wrote: > I've noticed that 'make reindent' creates big patch files too, it is > fixed in svn revision 1495 (by invoking indent twice). > > Attached you'll find the simplified version.
Thanks. > > > (In the meantime I've released 0.12.3, but a 0.12.4 could follow soon, > > > if required) > > > > The patch does add a new string (which shouldn't be seen, but should be > > translated), so that will need a string-freeze waiting period. > > String freezes will be no problem, there are already modified strings > for 0.12.4. > > A release in Debian of 0.12.3 would be nice (ergo not waiting for > 0.12.4), because many users would then be able to play games with a > server on localhost again. Ok, I'll try to release that soon then. > Now about the requested functionality for IPv6 support. > It is not clear to me what the required behaviour would need to be > (regarding Debian). What should the server do when either the IPv4 or > IPv6 socket is already in use? I would recommend that the server will > refuse to be run at all, otherwise situations would exist where the > client would connect with the wrong protocol to another listening > process, see the scheme below. I agree. Listening on one protocol and not the other is a bad idea. It only leads to confusion. The user will expect other programs to listen on both protocols as well, so when one of the is taken, "the port" is in use. > Situation 4: > * The log gets a line as in situation 2. > * Two warnings as in situation 3 are sent to the console. > * When the 'Stop Server' button is pressed, another line is sent to the > log window 'Preparing game...' followed by a line as in situation 2, and > a warning as in situation 3 is sent to the console. > It looks like the server is started twice, the first time with two > protocols and the second time with one protocol. That looks like a bug. > I would prefer a situation similar to situation 2. This seems like a bug, indeed. It must be fixed before the patch is included. > About the code: I would recommend to use GSList (or something similar) > instead of int* and a counter variable. Good idea. Thanks, Bas
signature.asc
Description: Digital signature
