André Warnier wrote: > Or, take this another way : suppose the proxy server isn't there, and > the client connects directly to the final server. What do you think > happens then, that is any different ? > If the client goes away while the server is busy generating the response > (say, consulting a database), the server will also notice only when it > tries to send something to the client, through the now useless socket. And so, it can in this case rollback a transaction, for example. But if the perceived client, mod_proxy, is still there listening for its answer, it can not do that and mistakenly assumes the client received it. > Or, yet another way : there is no "interrupt" generated when the client > goes away, which would enable the server to notice the fact, before it > tries to write to, or read from, the client socket But it could check it, at least immediately before starting to write the answer back, specially if it could have taken a relatively long time processing the request (a long order with many itens, for instance), couldn´t it? (That´s what a WebService I wrote is now doing as the language used was accepting some writes to the socket already closed by the client without generating errors which worked nicely; but then I faced this mod_proxy annoying behavior). ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. --------------------------------------------------------------------- The official User-To-User support forum of the Apache HTTP Server Project. See <URL:http://httpd.apache.org/userslist.html> for more info. To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxx " from the digest: users-digest-unsubscribe@xxxxxxxxxxxxxxxx For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxx
|