Gmail's IMAP implementation implements three RFCs I wrote; I haven't noticed anything wrong in those areas. Are my RFCs unusual or am I overlooking something?
Gmail's IMAP implementation implements three RFCs I wrote; I haven't noticed anything wrong in those areas. Are my RFCs unusual or am I overlooking something?
A sibling comment has already mentioned the folder/flags mixup issue, but while it sounds like one small feature, that one thing is so fundamentally broken that I really can't see how any user of Gmail IMAP would overlook it.
I get duplicate notifications for the same mail in INBOX and [Gmail]/All Mail and actions on one copy don't affect the other unless clients implement Gmail-specific hacks (which many/most do).
What RFCs are you referring to?
[0] https://bugzilla.mozilla.org/showdependencytree.cgi?id=40279...
The gp was talking about RFCs and the IMAP protocol implementation itself, so I was talking about the Gmail server at the protocol level, rather than user interface.
In addition, many Gmail features, especially security-related features, don't work with Mail.app.
That is absolutely a problem I've had with Mail and Gmail. No idea how I fixed it. Or whether it's been fixed at all. But for some reason, drafts turned up in searches, which made searching through my mail an incredibly painful experience. Search is fine now, though. Maybe I turned IMAP drafts off; I really don't know.
Showing the gmail labels as flags would be legal by the protocol (and have some disadvantages), showing them as mailboxes also legal (and have some other disadvantages), so I don't count that as a bug, it's merely a situation where one has to choose disadvantages.
The duplicate notifications are different; the way IMAP is designed the server cannot know which TCP connections originate from the same client. It can guess based on IP address and some other things, and it can know something based on RFC 2971, but 2971 also says explicitly that the server shouldn't do anyhting differently based on that. Which has been discussed in one of the IETF WGs, and the consensus was that the server isn't permitted to suppress probably superfluous notifications. If a client opens two connections, getting rid of duplicate notifications is the client's job.
www.rfc-editor.org/search/rfc_search_detail.php?author=gulbrandsen are my RFCs; I actually wrote most of those and gmail implements four of the IMAP-related ones I've written.
There's no line in the IMAP spec. that very explicitly states "each mail MUST exist in exactly one folder", so you can certainly argue that Gmail technically follows the spec. but the only reason that line doesn't exist is because doing otherwise is so non-sensical as to not have been considered as a possibility. It does for example contain the following line, which Gmail does not follow: "permits manipulation of mailboxes (remote message folders) in a way that is functionally equivalent to local folders", but the fact that line is simply in the abstract and contains no MUST means one can argue it's not prescriptive.
There may be disadvantages to using flags instead of mailboxes, but equating their disadvantages is reverse hyperbole. If there was any equivalence, you wouldn't have every major client containing a rake of Gmail-specific code to work around the issue.
I don't want to argue about the disadvantages of flags vs. mailboxes. It's a valid subject but I don't want a subject change.
There are advantages and disadvantages to both approaches; it must have been a nasty choice to make.
Sent emails for one.
I’ve written various email integration components and have to make special safe-guards for Gmail-specific crap.