In message <[EMAIL PROTECTED]> Sam Varshavchik <[EMAIL PROTECTED]> wrote:
> Although I will correct UIDNEXT, your method for checking of new mail is > wrong. A UIDNEXT comparison would only be valid if the folder's UIDVALIDITY > remained the same, which you fail to check for. Good point. I was somewhat aware that I was gambling on that but my problem is that I am gatewaying to another protocol and there isn't anything much that I can do if the UIDVALIDITY changes :-( That trace I supplied was actually from a slightly out of date version as the latest ones also fetch the UNSEEN count and pass that on via the other protocol as that now has support for a count of unseen messages. Unfortunately in seems that the client program that my gateway talks to still prefers to use the high water mark information to guage when new message arrive so I guess I will have to talk the author about that. > Furthermore, RFC 2060 does not require the folder's UIDVALIDITY to remain > constant all the time. A compliant implementation could be, for example, > UIDVALIDITY being simply the current system time, with all messages in the > folder always having a monotonically increasing UID starting at 1. Whilst that would certainly appear to be conforming I think it would be likely to seriously degrade the performance of many clients. > In short, this is not the correct way to check for new mail anyway. I wish I knew a good and efficient way to check a number of folders for new mail... I have tried a number of ways over the years. Tom -- Tom Hughes ([EMAIL PROTECTED]) http://www.compton.nu/ _______________________________________________ courier-users mailing list [EMAIL PROTECTED] Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
