Sorry Lennart, but you are wrong once again

Lennart Poettering’s post in G+ is gathering a lot of attention these days, most of the feedback is supportive, and positive, which is not surprising to me, because although Poettering would like us to believe otherwise, most of the open source community is pretty accommodating and non-confrontational.

I am however going to go against the current here, and criticize him, but first let me state clearly that I do not condone any physical attacks towards his person, or the threats of such. His ideas however are a different matter.

Lennart’s chief mistake is to attack the way the Linux’s kernel community is run, and say their success happens despite this. How does he know? Has he ever run a more successful community? Has anybody ever? Linux is the most successful software project in history, by more than one order of magnitude from any way you look at it. It would be presumptuous for anybody to say they know how to run this project better, specially without any evidence to back such claim, which is precisely what Poettering is doing.

In this blog I’ve analyzed the many reasons why the Linux kernel is so successful, and one of them is its combative style of discussion in which ideas are not exempt from ridicule, and strong language is often used to drive one’s point home as efficiently as possible. Many people in the community agree this is desirable, and there’s even scientific evidence that supports this notion; the best ideas arise in a confrontational environment, not in a protective one.

What’s more, Poettering himself accepts he hasn’t been involved in this community. So what the hell does he know about it? Nothing.

Poettering’s second mistake is to assume that for non-white, non-western, non-straight people the situation surely must be worst… That is not the case. Maybe, just maybe, he receives such vitriolic feedback not just because of what he does, but because of the horrible way he does it. Of course not, Poettering doesn’t need to change, his approach is perfect, in fact, the only reason he receives criticism is because he is too progressive, too audacious, too efficient, surely, that must be the reason!

Personally, my beef with Poettering starts from the fact that he blocked me from Google+. Why? Because I was complaining about a technical issue with systemd, which he initially spotted and commented, but then ignored. In the middle of the discussion I made some value judgements about certain systemd code, and he stopped responding and blocked me. That is the worst way to end a discussion; block the people who disagree with you.

Sorry Lennart, but actions have consequences, and you can only do so much disruptive changes to the Linux ecosystem without much care or consideration for others, there’s a limit to the amount of people you can block, and the criticism you ignore. You can grow as thick a skin as you want, you are still wrong. No community is going to let you continue being wrong and acting as if you are beyond reproach just like that (unless you run that community and have blocked any dissident voices of course).

Maybe it’s time to take a hard look in the mirror.

Advertisements

Avoid Reply-To munging; mail as mail was meant to be

Reply-To munging (the act of overriding the ‘To’ header in mails) is a common practice on many mailing lists, but does it provide any value?

I participated in a long and tedious discussion on fedora-devel, not only regarding Reply-To munging, but also allowing non-subscribers to post. At the end I realized why mailing lists do such a thing: inertia, misconceptions, and laziness.

After shutting down all the arguments provided, it turns out there’s only one benefit of Reply-To munging; I’ll try to explain why it’s not so important, all the problems it causes, and some of the invalid arguments.

Personal level indicators

Before getting started I want to show a cool feature of Gmail that is only possible with proper replies (unmunged).

Personal level indicators

With these it’s very easy to see which mails has been addressed to me personally while browsing a mailing list. For this to work, of course mails needs to be addressed to me (like it normally is), instead of the mailing list. And of course Gmail is not the only client that supports this; mutt can highlight mail with the ~p filter.

There are similar advantages, like the fact that I can search for “to:me”.

In favor

Virtually all mail clients provide at least two options when replying mail:

  1. Reply to sender
  2. Reply to all

Sometimes people mistakenly click “reply to sender”, but when Reply-To munging is used; the sender is the mailing list, so when people make this mistake (it still is a mistake), the mail still reaches the mailing list.

Everybody is aware of “reply to all”

It might come to a surprise to many hard-core mailing list users, but the common (non-geek) population of e-mail users is perfectly aware of the “reply to all” button; otherwise they wouldn’t be able to have group conversations.

People have back-and-forth conversations to more than one person through e-mail; therefore people know when to use “reply to all”. So the argument should not be on ignorance of the option, but mistakingly using the wrong one.

Gmail, Outlook, Hotmail, Thunderbird, Yahoo! Mail; they all have the option, because people use it.

It still is a mistake

Many people think that if the mailing list has Reply-To munging, then you can stop using “reply to all”; that’s not true.

The annoying “Please keep me in Cc as I’m not in the mailing list” warnings would not be needed if people clicked “reply to all” (even when the To header is munged); the munging applies to the To header, but not the Cc headers.

So for example, I Cc myself on munged mailing lists; if people use their mail clients properly, then I’d be addressed properly and have all the advantages of that, like personal mail notifications. But no, most people are lazy and click “reply to sender”.

Munging only perpetrates this bad behavior.

We don’t need no babysitters

Would you like the mailing list to bounce messages back saying “it looks like you forgot to add the attachment file”? This mistake, just like clicking the wrong reply option, is something the user should avoid doing.

Against

The previous argument “in favor” is very week at best, but there’s still some value in it. Is it really worth it when measuring the arguments against?

Personal level indicators don’t work

I already explained this feature is very useful, and if it wasn’t clear at this point: munging makes it impossible.

A few people do use “reply to all” even on munged mailing list, but they are the minority, the fact of the matter is that you can’t rely on this: when you see a message without a marker it might still be addressed to you, so you can’t dismiss it as quickly as you would on a non-munged mailing list.

Impossible to do “to:me” searches and filters

Quite often when searching mails you want one of the criteria to be “addressed to me”, however, if the To field was munged, then you just can’t do that, you would have to reply on full-text searches, or other sub-optimal method.

Some clients support tags (like notmuch), so it’s possible to say: hey, put all mail addressed to me on the inbox. Munging prevents that.

IETF is clear

I found this one from here.

In April of 2001, the IETF issued af new document, RFC 2822, which obsoletes RFC 822. In this new RFC, the author addresses the Reply-To header field in a few places, but the most relevant to this discussion is the following in section 3.6.2 “Originator fields”:

When the “Reply-To:” field is present, it indicates the mailbox(es) to which the author of the message suggests that replies be sent.

Your list software is not “the author of the message”, so it must not set or in any way meddle with the Reply-To header field. That field exists for the author and the author alone. If your list munges it, you are violating the standard.

Invalid arguments

I think I’ve mentioned all the important arguments, but some people keep bringing these, which are obviously invalid.

Double mails

Some people argue that if Reply-To munging prevents the same mail to be delivered twice to the sender (one directly, and one through the mailing list). This is of course nonsense because the mailing list allows you to decide whether you want the software to send you the copy, or not (since the software can see you are on the list of recipients).

Not to mention that you can add the Reply-To header if that’s the behavior you want; there’s no need to punish the rest of us.

Useful?

There’s a popular rebuttal to the famous ‘Reply-To’ Munging Considered Harmful by Chip Rosenthal; Reply-To Munging Considered Useful.

RFC 822 and “Text Message Teleconferencing”

Obsoleted by RFC 2822.

The Principle of Minimal Bandwidth

There’s no “double mail” problem, so there’s no extra bandwith.

Reply-To Munging Adds Something

Very few people want to send mails only to the list, and they can do the same than the people that want to reduce the Cc list: manually edit the fields. Not to mention that some clients have the “reply to list” option.

There’s no double mail, there’s no snowball.

It Doesn’t Break Reasonable Mailers

Nonsense; the functionality being broken is not replying only to the author, but replying to the full To list (and Cc list).

Again nonsense; new options can be achieved for a minority of users. The decision of what the mailing list does should rely on the behavior of mailing list as a whole, for which only the only thing that matters is the majority of mailers right now.

Freedom of Choice

Wrong. The option to reply to all the recipients is taken away. The ability to determine to whom the mail is actually addressed is also gone.

And in fact this argument is shooting itself in the foot; if the mailing list doesn’t munge, then the user can choose whether to munge or not, but if the mailing list munges; the user has no option.

Some Mailers are Broken

Not true. All applications have “reply to sender” and “reply to all”.

Principle of Least Total Work

This assumes the principle of “Minimal Bandwidth” is true, which is not, and people would want to follow it, which they don’t. So 90% of the time (according to that argument) you want to reply to all the recipients (including the mailing list), and 10% of the time you want to reply only to the author. Both cases can be achieved with the usual options “reply to sender”, and “reply to all”.

So this argument is again shooting itself in the foot; the 10% of time where you want to reply only to the author is actually more work when munging.

People are Responsible for Their Own Mistakes

That counter-argument is correct, but I’m not arguing that Reply-To causes people to wrongly send personal e-mails to the list.

And in the End…

Funny, right after arguing that people are responsible of their own mistakes, an argument that the mailing list should prevent mistakes of continuing a thread off-the-record. It’s policy that should keep mails on the list.

It’s What People Want

This is a value judgement, not an argument. People change opinions, mailing lists changes members, and people often don’t know what the really want.

For more invalid arguments, check my summary on fedora-user mailing list.

Conclusion

So, from my perspective it’s pretty clear that Reply-To is something for mail authors to set, not mailing lists. It prevents essential features from working, and provides virtually no value.

The arguments are crystal clear, and putting old misconceptions aside, only stubborn morons wouldn’t be able to realize that Reply-To munging is actually bad. Right?