From dlevin at cs.tu-berlin.de Sun Aug 3 16:06:16 2008 From: dlevin at cs.tu-berlin.de (Dan Levin) Date: Sun, 3 Aug 2008 22:06:16 +0200 Subject: [OTR-users] otr reliable, in-order messages Message-ID: <308E4266-37C4-4BD6-BDAF-BDBE39E6B12A@cs.tu-berlin.de> Hi fellow OTR users/devs, Are there any plans to explore or implement guaranteed reliable, in- order message delivery in OTR? This would be a very valuable feature both from a security and reliability standpoint. If anyone knows of any work already done in this area, please let me know, otherwise any thoughts or comments would be welcome. Thanks! -Dan -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 2427 bytes Desc: not available URL: From ian at cypherpunks.ca Mon Aug 4 15:25:37 2008 From: ian at cypherpunks.ca (Ian Goldberg) Date: Mon, 4 Aug 2008 15:25:37 -0400 Subject: [OTR-users] otr reliable, in-order messages In-Reply-To: <308E4266-37C4-4BD6-BDAF-BDBE39E6B12A@cs.tu-berlin.de> References: <308E4266-37C4-4BD6-BDAF-BDBE39E6B12A@cs.tu-berlin.de> Message-ID: <20080804192537.GZ18985@yoink.cs.uwaterloo.ca> On Sun, Aug 03, 2008 at 10:06:16PM +0200, Dan Levin wrote: > Hi fellow OTR users/devs, > > Are there any plans to explore or implement guaranteed reliable, in- > order message delivery in OTR? This would be a very valuable feature > both from a security and reliability standpoint. If anyone knows of > any work already done in this area, please let me know, otherwise any > thoughts or comments would be welcome. OTR leaves that up to the underlying IM network. OTR assumes the network already provides in-order delivery (doing IM when messages can get reordered seems like something no IM provider would allow), but allows for messages to be dropped (one party goes offline, messages are dropped, etc.). I don't think guaranteed delivery is even something we *want*, since if your buddy is offline, that would imply there's a long-lived decryption key somewhere, and that's defninitely undesirable. - Ian From gmaxwell at gmail.com Mon Aug 4 15:38:07 2008 From: gmaxwell at gmail.com (Gregory Maxwell) Date: Mon, 4 Aug 2008 15:38:07 -0400 Subject: [OTR-users] otr reliable, in-order messages In-Reply-To: <20080804192537.GZ18985@yoink.cs.uwaterloo.ca> References: <308E4266-37C4-4BD6-BDAF-BDBE39E6B12A@cs.tu-berlin.de> <20080804192537.GZ18985@yoink.cs.uwaterloo.ca> Message-ID: On Mon, Aug 4, 2008 at 3:25 PM, Ian Goldberg wrote: > On Sun, Aug 03, 2008 at 10:06:16PM +0200, Dan Levin wrote: >> Hi fellow OTR users/devs, >> >> Are there any plans to explore or implement guaranteed reliable, in- >> order message delivery in OTR? This would be a very valuable feature >> both from a security and reliability standpoint. If anyone knows of >> any work already done in this area, please let me know, otherwise any >> thoughts or comments would be welcome. > > OTR leaves that up to the underlying IM network. OTR assumes the > network already provides in-order delivery (doing IM when messages can > get reordered seems like something no IM provider would allow), but > allows for messages to be dropped (one party goes offline, messages are > dropped, etc.). > > I don't think guaranteed delivery is even something we *want*, since if > your buddy is offline, that would imply there's a long-lived decryption > key somewhere, and that's defninitely undesirable. OTR does already provide a useful property along these lines: The ability to get something like a negative acknowledgment. A problem I have with some of my AIM contacts is that they'll fall offline but appear to be online for some time. By smacking refresh connection I can see if their client is still around and differentiate the human not responding case from the computer disconnected one. Along those lines it might not be too unreasonable to make OTR communicate back ID numbers of the last several received messages anytime you send a message or refresh. Then your client would learn which of your prior messages didn't make it. I wouldn't go as far as suggesting retransmission, but knowing which messages aren't going through could be an enormous confusion reducer on unreliable links.