On Sun, 9 Sep 2001 13:36, Guus Sliepen wrote: > On Sun, Sep 09, 2001 at 01:31:14PM +0200, Russell Coker wrote: > > > Actually, one good solution is turning on the Nagle algorithm, or > > > rather not using TCP_NODELAY (I don't know if ssh sets it, but I guess > > > it does to improve responsiveness). What you propose will break > > > situations where you have to type one non-interesting key to invoke an > > > action, people wait for > > > > No, the server just registers that key as a key that will push all the > > data. It would be more efficient than anything that the Nagle algorithm > > could produce AND give the best responsiveness. > > Well, Nagle is very good for fast typers :). > > > BTW All correctly written terminal programs will push all data out and > > defeat the Nagle algorithm anyway. > > I don't know whether that really is correct behaviour. Depends on your > point of view I think.
Doesn't Nagle typically involve delays of >100ms? >100ms extra delay for a serial link in unacceptable for interactive use, so every key press must be pushed. -- http://www.coker.com.au/bonnie++/ Bonnie++ hard drive benchmark http://www.coker.com.au/postal/ Postal SMTP/POP benchmark http://www.coker.com.au/projects.html Projects I am working on http://www.coker.com.au/~russell/ My home page