haystack / murmur

A mailing list designed to reduce noise and encourage sharing
25 stars 13 forks source link

sometimes replies become a brand new post instead of being added to thread? #83

Open amyxzhang opened 8 years ago

amyxzhang commented 8 years ago

This doesn't happen that often but occasionally does, and I'm not sure why.

karger commented 8 years ago

May have to do with failure to recognize a message-id or in-reply-to id that would make clear it is associated. Possible that some clients are not sending it. Are you using such IDs or just trying to match on the subject line (which has its own errors, including turning independent messages into replies)

On 04/21/2016 04:50 PM, Amy Zhang wrote:

This doesn't happen that often but occasionally does, and I'm not sure why.

— You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83

amyxzhang commented 8 years ago

We are matching via subject line alone (this is what I read somewhere that gmail does as well) and looking for a "Re:" before the subject.

Could do something smarter (such as checking dates or reply ids) in addition.

On Thu, Apr 21, 2016 at 4:52 PM, David Karger notifications@github.com wrote:

May have to do with failure to recognize a message-id or in-reply-to id that would make clear it is associated. Possible that some clients are not sending it. Are you using such IDs or just trying to match on the subject line (which has its own errors, including turning independent messages into replies)

On 04/21/2016 04:50 PM, Amy Zhang wrote:

This doesn't happen that often but occasionally does, and I'm not sure why.

— You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83

— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83#issuecomment-213109623

Amy X. Zhang | Ph.D. student at MIT CSAIL | http://people.csail.mit.edu/axz | @amyxzh

karger commented 8 years ago

So if someone changes the subject line you won't match. Maybe that's what's happening. I'd recommending using suitable id matching when possible, as it is more reliable when it exists.

On 04/21/2016 04:55 PM, Amy Zhang wrote:

We are matching via subject line alone (this is what I read somewhere that gmail does as well) and looking for a "Re:" before the subject.

Could do something smarter (such as checking dates or reply ids) in addition.

On Thu, Apr 21, 2016 at 4:52 PM, David Karger notifications@github.com wrote:

May have to do with failure to recognize a message-id or in-reply-to id that would make clear it is associated. Possible that some clients are not sending it. Are you using such IDs or just trying to match on the subject line (which has its own errors, including turning independent messages into replies)

On 04/21/2016 04:50 PM, Amy Zhang wrote:

This doesn't happen that often but occasionally does, and I'm not sure why.

— You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83

— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83#issuecomment-213109623

Amy X. Zhang | Ph.D. student at MIT CSAIL | http://people.csail.mit.edu/axz | @amyxzh

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83#issuecomment-213110470

kmahar commented 8 years ago

So I'm looking at headers from some Murmur emails now, and it looks like there is a problem with the way we are setting message IDs, because Google keeps overwriting our IDs with things like 5718049e.848c6b0a.a5708.335fSMTPIN_ADDED_BROKEN@mx.google.com and saving the ID we set as "X-Google-Original-Message-ID". I think this may because the header we're setting doesn't surround the message ID with < >'s. I will test it out.

Each thread seems to gather a list of "references" in a header that are previous message IDs from the thread. So it seems like we could properly set a message ID on the original post and save it along with the thread, and then that would hopefully always be in the "references" header of all emails sent later on; and then we could match using that.

karger commented 8 years ago

https://tools.ietf.org/html/rfc4021#section-2.1.9

On 04/21/2016 04:55 PM, Amy Zhang wrote:

We are matching via subject line alone (this is what I read somewhere that gmail does as well) and looking for a "Re:" before the subject.

Could do something smarter (such as checking dates or reply ids) in addition.

On Thu, Apr 21, 2016 at 4:52 PM, David Karger notifications@github.com wrote:

May have to do with failure to recognize a message-id or in-reply-to id that would make clear it is associated. Possible that some clients are not sending it. Are you using such IDs or just trying to match on the subject line (which has its own errors, including turning independent messages into replies)

On 04/21/2016 04:50 PM, Amy Zhang wrote:

This doesn't happen that often but occasionally does, and I'm not sure why.

— You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83

— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83#issuecomment-213109623

Amy X. Zhang | Ph.D. student at MIT CSAIL | http://people.csail.mit.edu/axz | @amyxzh

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/haystack/murmur/issues/83#issuecomment-213110470

kmahar commented 8 years ago

oh yeah - that seems simpler.