-1
> -----Original Message----- > From: Sheng Yang [mailto:sh...@yasker.org] > Sent: Thursday, February 07, 2013 4:01 PM > To: Chip Childers; Alex Huang; Brett Porter; Animesh Chaturvedi; David Nalley; > Edison Su; run...@gmail.com; dk...@apache.org; > htrippa...@schubergphilis.com; shadow...@gmail.com; > somikbeh...@vmware.com; Frank Zhang; w...@widodh.nl > Cc: cloudstack-dev@incubator.apache.org > Subject: [VOTE] Revert back to old mailing list mechanism, which would add > "Reply-To: mailing list" to every mail it send out > > Hi all, > > I'd like to call for a vote for reverting back to the old mailing list > mechanism, > which would add "Reply-To: mailing list" to every mail it send out. > > And I need to declare that I would vote *-1* on this revert. > > Whatever you voted in the previous mail, I suggested to read the whole mail > before vote. > > Here are some backgrounds: > > 1. What's "Reply-To" header > > Defined by IETF RFC 5322(the latest version of "Internet Message > Format")[1], 3.6.2 Originator Fields: > > <quote> > When the "Reply-To:" field is present, it > indicates the address(es) to which the author of the message suggests > that replies be sent. > </quote> > > Which means, this option would override the default behavior of replying > mail, to send out mail to the specified mailing address (mailing list address > in > this case) rather than original author of the mail. > > 2. What's the old mailing list mechanism > > Long ago, many people familiar with other mailing list like LKML or libvirt > realized there is no way to use reply all to the author and this mailing list > as > we did before on this mailing list. The mail only goes for the mailing list > address, not for the author. That's because in the past, this mailing > list(cloudstack-dev) added "Reply-To" field to all the mail it sent out, which > would override the original author field when others reply the mail. So > something like this would happen: > > Event: A wrote mail X, send to mailing list. > Result: X reached other subscribers of mailing list, with "From: A" > and "Reply-To: M" (mailing list). > Event: B replied the mail X. > Result: X reached other subscribers of mailing list, with "From: B" > and "Reply-To: M". There is no A mentioned in this mail's header. A would > have to check the mail from mailing list to know B replied. > > 3. What's the new mailing list mechanism(which is happening now). > > The "Reply-To" has been discard. So every mail come along would go back to > it's author as well as the mailing list. > > Event: A wrote mail X, send to mailing list. > Result: X reached other subscribers of mailing list, with "From: A" and "CC: > M". > Event: B replied the mail X. > Result: X reached other subscribers of mailing list, as well as A's mail box > directly, with "From: B" and "To/CC: A, M". A would see that in his inbox > directly. > > 4. What's the pro/con of the old approach(I won't vote for this, so you know > this may be bias). : > > Pros: > a. Enforcement: It would enforce every communication happened in the > mailing list. > b. Fix the broken mail client: You don't need to have a mail client support > "Reply-to-all" for involving the mailing list. > > Cons: > a. Violate RFC 5322. RFC 5322 said clearly that ONLY "author" can suggest to > use "Reply-To" for an alternative address of receiving the reply. Mailing list > server is NOT the author of the mail. > b. Inefficient: Everyone would setup a filter for mailing list would need to > dig > the mailing list from time to time to see if there is a response. > > And here is an very old article on explaining why "Reply-To" is bad thing to > do[2]. You can read if you're interested in. > > 5. What's pro/cons of the new approach: > > Pros: > a. Efficient: Author would receive the mail addressed to his mailbox, so he > would know that's a reply(from the mailing list) immedately. > b. Consistent: When you in the thread for multiple people, you won't expect > "Reply" single would reach all the people. That's why most people always use > "Reply-to-All" by default in their daily life. > c. Keep people in thread. Even if you're at a long weekend and don't like to > be bothered by mailing list but someone replied you on one month old > thread, you would know immediately. > d. More involving: People don't need to subscribe to the mailing list to > involve. Like Wido pointed out, most mailing list is doing this because they > encourage the anticipating, even temporarily. You don't need to subscribe to > the mailing list to involve in the community, but you still can choose to do > so > if you think it's good enough for subscribing. > > Cons: > New comer's mistake: It happened when one just begin the community life. > Someday he hit "Reply" rather than "Reply-to-all" by mistake. > Then mail didn't go to the mailing list. > > 6. My opinion: > > a. Inefficient is unacceptable. I don't want to spend any unnecessary time to > look through all the mails to find out what's my interested in, especially > when > I am in a tiger team and had worked for more than > 12 hours a day. > > b. Man made mistakes, but they learned quickly after that. I've learned that > as well. In fact I suppose most people would use "Reply-to-All" in the > company or daily life, so I don't think it's hard. Anyway, I set > "Reply-to-All" by > default in all my mail clients, and I expected most of us have done the same. > > c. Some people said it would encourage offline discussion. I distaste this > thought most. It seems you shouldn't been given freedom to choose > because we didn't trust you can do the right thing. But it's the trust which > build the community, and it's the freedom all Open Source/Free Software > about. "Free as in freedom". Yes, this approach just make it easier for people > to discuss offline, but does it matter? If you don't trust the people would > able to do the right thing, I am afraid even if you tried every method you > have to enforce it, they won't help a bit. > Community is about people, not about the mailing list. Offline discuss can > always happen if people want. Community is an spontaneously organization, > not an prison, or Soviet Union. People have right to choose. If you cannot > believe they would do the right thing if you give them choice, then this open > source community is already done. The Linux kernel mailing list or xen-devel > or kvm-devel or libvirt or many other famous mailing list, do it in this way, > and > none of them hurts because of "encouraging offline discussion". > > I vote -1 on this change. > > [1]http://tools.ietf.org/html/rfc5322 > [2]http://www.unicom.com/pw/reply-to-harmful.html > > --Sheng
