Paul J Stevens <[EMAIL PROTECTED]> said: > Aaron Stone wrote:
>> Are you suggesting that we would check the To and CC fields against a >> user's aliases, and only send vacations for things which appear to be >> "personally adressed" to the recipient? > > My point exactly. Either this was a useful clarification, or I read your reply to Udo to quickly... in any event, let's step back a second: Sieve can provide vacation support, and can do it in a [draft] RFC manner (the managesieve protocol, which is reasonably well supported). libSieve's vacation mechanism requires certain information from the calling application, and hands back a suggested vacation response, which the caller can choose to send if the right criteria are met. There's a struct in libSieve, (which will probably be replaced with a chattier function that takes more flexible arguments at some point), here it is: typedef struct sieve2_vacation_context { int min_response; /* 0 -> defaults to 3 */ int max_response; /* 0 -> defaults to 90 */ /* In Sieve 2, send_response is right here */ struct { char *addr; char *fromaddr; char *subj; char *msg; int mime; } send; /* In Sieve 2, check_response is right here */ struct { unsigned char *hash; int len; int days; } check; } sieve2_vacation_context_t; If we can build the foundation of vacation support that provides the necessary checks for whether or not a message should get a vacation reply, then we can plug it right into libSieve and put together a big piece of the sieve puzzle. Aaron --