Don’t call my baby ugly, and other Mark Crispinisms
In this article in Wired, Mark Crispin laments:
“I am very pleased that Gmail intends to adopt IMAP,” he says. (Note his word choice: “intends.”) “I feel that their current server should be considered to be a ‘work in progress’ and not as a viable ‘ready for prime time’ IMAP server.”
Crispin says if he were to rate Google’s current implementation of IMAP, it would be “quite damning.”
My question would be: Is there any implementation that isn’t horrible in Mark Crispin’s hallowed opinion? Besides the one he wrote, of course.
It seems to me that IMAP’s adoption rate suffers from two major issues:
- Requires an always-on connection to be of full value. My experience with caching clients for off-line activity is not overwhelmingly positive.
- The spec is gigantic.
Let’s look at the second. The first IMAP RFC was issued in 1988. That’s almost 20 years ago. Since then, there’s been 3 major revisions of the protocol, and nearly 40 additional RFCs issued around it. Not exactly “lightweight” in any definition that I can imagine. Heck, they’ve issued 3 extensions this year. Maybe that’s part of why it’s not been implemented “correctly”.
POP sucks, don’t get me wrong. But it’s been a pretty stable and simple protocol. IMAP, not so much. Mr. Crispin can whine about how his “baby” isn’t being treated right, but in the end, Google implementing 1/10th of the IMAP protocol will bring it to more people than any perfect implementation before it. Additionally, we’ll find out what really matters in the protocol. IMAP lost long, long ago to MAPI for client/server communication protocols. This may be a chance to salvage it, but not if people are more concerned with ideological purity than functionality and support.
Bonus points for putting Mark Crispin in the woods with a hunting rifle.
That certainly makes a “point”.This entry was posted at 10:02 am on 27 October 2007 and is filed under Technology. You can follow any responses to this entry through the post-specific RSS 2.0 feed.
Thunderbird and Evolution work fine offline, too. Never run into a problem with offline support for MAPI.
MAPI’s been dead for a while. Where have you been working that IMAP “lost” to MAPI?
Maybe this is why Microsoft never follow standards? They all sucks and have less functions
LOL – MAPI “dead for a while”? Last I checked, Exchange has complete and total dominance in the enterprise, medium, and small business spaces. IMAP didn’t just lose, it got utterly stomped.
First, Microsoft doesn’t follow standards for, I believe, two reasons. One, it allows market dominance when they control the de facto standard. Second, often the standards do not actually solve the problem they’re trying to solve, and their underlying driver won’t let them participate in fixing the standard.
If I were to extrapolate from my experience in small, medium, large businesses and government, I would say that MAPI represents upwards of 90% of the market. Perhaps it doesn’t represent the hobbyist very well, nor does it represent certain organizations, but the reality is quite different than the perception from some corners. MAPI sucks, having had to reverse engineer large chunks of it, but it does work. Mostly.
As for offline access, my experience with Mail.app, Thunderbird, and others, is that they often lose sync with things, and status, and that resyncing often causes emails to disappear and appear that have long since been dealt with. Now, is this the client’s fault? Perhaps, perhaps not, but people don’t accept that “oh, it’s the server’s fault”. It’s the solution’s fault, and the combination of server and clients that I’ve had to work with (including Mark Crispin’s work) don’t work smoothly in a lot of cases of disconnected work.
Both comments and pings are currently closed.
To name just one program, Mail.app is very good at offline access. Messages are downloaded in their entirety, not just headers, and local copies can be kept of everything. I can send emails offline and have them sit in an outbox until my laptop gets WiFi.
If you’re going to say that IMAP is too complicated, compare it to a like protocol. You cite MAPI. Is MAPI simpler or manageable than IMAP? Which has more implementations? My understanding had always been that MAPI succeeded because of market forces, not technical ones.