Shared publicly  - 
 
So, FESCO's stone age fraction won, Fedora will install sendmail for another twenty releases as default. My feature to remove it from default failed, and what's particularly sad is that the deciding vote that caused it to fail was actually by the other person behind the feature, who didn't bother saying anything until the final vote where he simply didn't support it anymore. What a disappointment, both personally and technically.

I'd really like those three days of my life back that I spent on this idiotic fedora-devel flamefest last week.

If FESCO decides that Fedora's home is the stone age, then that's fine, I don't think I have to care anymore.  They should really drop the "F" for "First" though from their 4-F motto, it's a blunt lie. Fedora is seldom first on anything non-trivial, anymore. It is just another conservative distribution. Sendmail is just the pinnacle of it. I mean, really? sendmail???

I have fought more than enough fights on fedora-devel, I had some successes with my own stuff, but it's a pity how many missed other opportunities there are. The fact that Fedora doesn't actively participate in technologies like btrfs is just disappointing.

There were previous attempts to get Sendmail out of the default install. They all failed. Mine too. Let this be a lesson to anyone who thinks that 19 releases with Sendmail in the default are a tragicomedy, that needs to end. It's now a never-ending tragicomedy.

I for one am through with trying to change Fedora. I'll leave Fedora to the stone-age fraction. Thankfully there are always options to work around FESCO, And that's what I will focus on now.

Fedora certainly is the slowest booting distribution of all because of all the redundancy and cruft it ships. There's dire need of some champion to clean that up, reduce what is in the default boot. But that's not going to be me. My own machines do boot in < 4s including BIOS now, it was a nice dream to make a mainstream distro boot in similar times by default. I guess that dream is dead.

And now, I am going back to doing something useful. Like hacking.
103
13
Elad Alfassa's profile photoThomas Andersen's profile photoSirko Kemter's profile photoGimmeapill Eko's profile photo
59 comments
 
Why, integrate your own SMTP server into systemd ... ;-)
 
Didn't you know that Fedora's F for "First" was recently changed to "Floppy Support"?
 
I'm surprised sendmail was so controversial. Ok you need something that meets the LSB maill delivery agent if you install the lsb feature set and you might even be able to make a case of having it on disk but not activated from boot unless its turned off for receiving via SMTP or something runs the binary to send stuff.

Do you still ship vim and other such stuff ;-)
 
Been there, I know how that feels. After the Zif/yum flame-fest I've also fallen out of love with Fedora.
 
I'm a bit sad about it too, but I don't see this as the end of the world. It's out of 'core' now, perhaps next cycle we can nuke it from 'standard' as well. 
 
Lennart, I just want to thank you for all of your efforts over the years to fight this kind of stagnation and make the world a better place. It is really a shame that this failed. It seems so obvious. So many missed opportunities indeed. The development model of Fedora just doesn't work unfortunately.
 
+Alan Cox Well, Fedora isn't LSB compliant anyway. It only tries to be if you install the LSB package (which isn't in the default), and even then it doesn't really bother with certification or anything...

+Kevin Fenzi Well, "perhaps". But why would it result in anything different from the last three attempts? If 19 releases weren't enough to make people realize what ridiculous a choice sendmail in default is, why would this change within the next 6 months? Also, who will fight for that? Who has the time to spend another week on idiotic flamefests with retro-computing freaks on fedora-devel? And who wants this discussion every 6 months now?
 
I am curious to read the FESCo discussion, however I agree with +Kevin Fenzi, I think it was out of the window already but the @standard group rescued it. Try it again in 6 months and you will not be disappointed
 
+Kevin Fenzi you mean as was said there on the meeting when "fesco" feels like it. No FESCO voting against these features was based on their own personal preference and some incompetent administrators voices who apparently cant install a package into the distribution rather then some hard facts of technical limitation we might have to overcome.

If FESCO had done their due diligence and research it would already know about these two based on systemd+journal denyhosts and fail2ban replacements out there [1][2] so those arguments on devel where moot.

Any fallouts that might haven fallout of the rsyslog proposal  we could have caught with the earlier proposal which entire purpose was to do exactly that to prepare for the moment when we would eventually remove it and have identified and fixed or provided an alternative but FESCO does apparently not want people to work ahead of time in the community and now as they have shown they neither want to be inline with what other distribution have been doing nor for the project to follow one of it's four foundation.

And what Matthew did was indeed disgraceful

1. https://github.com/Lekensteyn/ssh-blocker
2. https://github.com/sofar/tallow
 
Postfix. The first thing I do on a new installation is to replace Sendmail with Postfix. 
 
Fedora participates in btrfs.  I'm not sure why you think it doesn't.
 
+Josh Boyer I guess the "First" mantra should tell us to push harder to have btrfs by default in the distro. I have been amicably trolled by friends for the failure of the corresponding proposed feature in several past releases.
 
+Gianluca Sforna pushing something to every user in Fedora before it's actually smart to do that isn't "First".  It's "Fail".

btrfs is making really good progress, and I think in perhaps a release or two it might be ready for use as the default FS.  Rushing it is just going to bring pain to everyone.
 
+Lennart Poettering I might give it a try for F21... if we can show how few people cared that it was gone from core perhaps that will turn the tide. 
 
This is sad. If Fedora is supposedly the playground for RHEL, then why the heck is sendmail still used? Every sysadmin I know immediately replaces it with postfix. At least replace sendmail with postfix I'd say.
 
+Ferry Huberts  The feature request was about removing it from core and standard not to replace it with something else or remove it entirely from the distribution....
 
Differentiation is key... floppy support is a key differentiator and a really smart move. Someone is going to kickstart a retro-computing tablet that looks just like a  5 1/4 inch floppy, with a slot for a 3 1/2 inch not-so-floppy disk.. and we will be there on day one..and it will be glorious.
 
+Josh Boyer Totally agree. However the damage was done, as many news sites picked it up and then we failed to deliver.
I hope we can still be First on this when the FS is ready.
 
Please don't quit.

You do a lot of boundary-pushing work in the OS.  This rubs some people the wrong way.  It's unfortunate that they won on this topic at this time, but what you are doing is too important to let a temporary setback completely derail you.  Not many people can do what you've been doing, particularly given the many difficulties.

I think you have to expect to lose a few.  If you look back your overall record is pretty amazing.
 
Do you have a list of such cruft you'd like to remove? It could be a nice start to make it easy to remove for those of us who do not need them, do not want to fiddle with the system but would like to testbed a lighter future...
 
serious comment. Have you thought about putting a slate of FESCO candidates together around a modernization mandate as a primary concern in FESCO election? Handling features is a big part of the job. I'd love to see a slate of candidates come forward who have some very specific opinions on where the featuring process should be taking the project.

Bringing change through brick and mortar governance structures have similar..hurdles. Failures of initiatives in the brick and mortar would can be the trigger which ushers in new candidates for governance.

I really don't think the modernizers have tried to make modernization important in how we elect our governance to grease the wheels of decision making.

Instead of quitting I would encourage you to change tactics slightly. I would encourage you to put together a slate of electable fesco candidates that will make a strong commitment to modernization. And sell that governanace platform and the slate of candidates... so I back them with my vote into a fesco seat... and hopefully lower the friction on modernization proposals.  
Allan Day
+
2
1
2
1
 
+Lennart Poettering - there are two tragedies here. The first is the decision itself, which would be hilarious if it wasn't so sad.

The second is that you had to spend so much time and effort trying to get this change through. No effective software outfit requires its most talented engineers to spend days on mailing lists dealing with bureaucracy. Successful organisations work by enabling their most talented to get shit done, not frustrating them and adding stop energy.
 
Wow. Luckily for RHEL we switched to a postfix default in 6. Hopefully they will kick Sendmail to EPEL in 7.
 
+Lennart Poettering so this debate is all about what comps group it's in (or not in), but that's not that relevant.  Spins have control over which groups they ship, and which packages they ship. We can fix the desktop spin to -sendmail and the other SIGs can fix their spins too.
We already modulate package list a bit to add nss-mdns, and get rid of system-config-printer.  I've edited the kickstart file now:

https://git.fedorahosted.org/cgit/spin-kickstarts.git/commit/?id=c79eede0d761d90a7a522ffc39420d9517136f0a
 
For the specific changes - removing functionality is by default not an improvement, and "Linux is an Open Source UNIX" is a description of the project as a lot of contributors joined it, so it's easy to see how preserving the UNIX aspects can be important to many.  It was fairly easy to guess that the proposals would be controversial and would require a lot of discussion even if they passed - so acting surprised that it took lot of time is IMHO not warranted.

Genrally re: bureaucracy - we have over a thousand contributors, so some level of bureaucracy is just unavoidable, and nobody gets everything they want (I certainly don't). "Working around FESCo" will just get everyone a more broken distribution and more unhappy contributors.

In general - again, removing things is not a program.  If we are not doing UNIX (and arguably we can't be "doing UNIX", any more), what are we building?  "Let's remove the broken things and build something better later" is not a a vision.  UNIX really has been one of the few common denominators - if you look at things that weren't in there, vast majority of them are fragmented into several independent and differing implementations, incompletely used and integrated (sure, with a few notable exceptions like libvirt).

One thing I'm really hoping for from the Fedora.next discussion is a serious exploration of the "integrated and designed OS" concept - beyond that one-sentence description there actually isn't any vision being proposed that we could follow.  (GNOME OS alone isn't enough because it doesn't account for the cloud and servers, not to mention that it the lower layers us from a 30-year old languages and tools to a ~20-year languages and tools, which is an improvement but probably not a sufficient one).  Now that discussion is easily worth a 1000-mail thread, as opposed from shaving a few megabytes from the default install.
 
+Jef Spaleta Well, the crux with that is that FESCO is actually a lot of work, they have weekly meetings and those meetings are sometimes long (like the one yesteday). If you are a member of the stone-age fraction of FESCO that time is easily available, after all you would never do anything new on your own anyway, that's why you are a member of the stone-age fraction after all. ;-) But for people who actually want to work on new stuff, that's some serious amount of time which one would have to spend which one would rather spend on something else.

I seriously dislike the politics of Open Source. All those folks who talk a lot and want to decide and never do anything. Those pointless bike-shedding discussions, were everybody has an opinion about trivialities but you only hear silence on the stuff that actually matters. Those self-sufficient people, for who only their own immediate use-case matters, who never bother with the big picture. Those conservatives for who religion, and folklore is a technical argument. People who have ensconced themselves in the status quo, who think the current mess that was Fedora was a great way to compete with the others. Those who think what we already have was good enough to win the rest, ignoring the irrelevance that Fedora is actually stuck in. People who have no thirst for making things better, but only the will to administer a legacy. People who think stop energy was a useful driving force in the computer industry. People who think the competition waits for us. People for who the 1970s haven't ended yet.

And you know what, I never wanted to be one of those guys. I always wanted to be somebody who does, and when he needs to also talks, but not one of those guys who just talks and never does.
 
"First they ignore you, then they laugh at you, then they fight you, then you win." :)
 
This is not about saving few megabytes, this is about removing unwanted service from being started on tens of thousands of desktops by default. By the way, Linux has never been UNIX.
 
+Ray Strode well, it's a work-around. I am the wrong guy for work-arounds. ;-)

Work-arounds are fine to make something work that just needs to work quickly. But we area actually designers, builders of something generic. It shouldn't need work-arounds.
 
As I just said in an e-mail to Lennart, I don't think this was a failure, or lack of progress. The decision is to move the package from mandatory @core to @standard. And, it specifically includes the note that we'll look at this again for the next release.

I decided to abstain from the part of the vote about the larger part of dropping it entirely, because while I'm absolutely favor of that, I'm respectful of the idea that the base design is important. I don't buy "must never change!" as an argument, but while I'm skeptical of "/usr/bin/sendmail is a useful modern API",  when people like +Miloslav Trmač are asking for a chance to articulate that as part of a real definition of the OS layer, I'm willing to listen. So, I abstained.

I think that Lennart ran out of the meeting after that, but I did vote for removing rsyslog from both minimal and the default, because the arguments there were less convincing to me. That ultimately got the same final vote, of remove from minimal and reevaluate next release, which is, again, progress, and gives us a chance to work on some of the things which are legitimate concerns here, like access to the logs from guestfish, fail2ban support for the journal, and refactored rsyslog packaging (with default /var/log/messages as a subpackage rather than always installed, so it's easier to drop into place as a remote-only logging solution).

To me, all of this is progress, and while we're not the first to make these changes, it's not like this is really blocking any new functionality from landing in Fedora. It's just cleanup. It's okay for that to be done carefully.

I'd love to see Fedora have a better overall story for alerting and log handling beyond local storage and viewing. The faster something like that can land, the easier it is to move away from the old thing.
 
As a Design Suite spin maintainer, I would like a list of legacy and depreciated packages so I can remove within ks file.
 Side note: it seems desktop team is taking care.
 
+Lennart Poettering
 Well, it takes a small group to make a major revolution. Failure is part of life and a process to view what went wrong, correct and come back stronger than ever. I will take that your proposal exposed a much inner issue within Fedora: conservatism.
Look at this way, you are by far a better game changer standing against a wave of resistance: pulseaudio and systemd are the biggest improvement within Linux ecosystem.
Think this as postponed plan.
 
+Lennart Poettering  I can understand not liking having to navigate the political and governance arena. I empathize. I also recognize that I'm probably not going to earn your esteem just by talking about what I would and would not like to see.

There absolutely is a conflict in prioritizes in the Fedora contributor base. Are the governance structures biased towards people interested in legacy compatibility? Maybe. Are modernizers more discouraged or self-selecting from being a part of the governance? Maybe. I've no reason to discount your personal testimony for why you don't want to be part of governance yourself.
But I've seen similar complaints of an opposite nature when big changes are pushed as well. So if everyone is basically unhappy with how things work... how do you make it work better for everyone?

I'm not seeing a doers approach to fixing the governance model outlined here.  I've made a suggestion on what I think modernizers can do to take more control...more ownership...of the process...but if taking the time to be part of governance isn't what you or other supporters of a faster pace of modernization isn't something you want to do.. I can't make you do it...and I'm not going to shame you for your reluctance either. Governance is a shit job.

What I see is frustration with the current decisions by governance populated with people who have different priorities for the project than you do. I see a stated lack of interest in standing up to taking on a governance role to ensure better decision making from a modernization and competitiveness pov. But I don't see any offered solution on how to fundamentally fix the governance model that is causing you the frustration.  The project must have some governance in place...as there established competing interests that most assuredly cannot all be simultaneously pleased with the resources on hand.

-jef
 
Eventually the stop energy that goes around on loads of mailing lists gets to you. Really annoying.
 
How about an edgy fork of Fedora.
We could call it .. uhm .. hoodie.
 
+Matthew Hannigan I would be interested in seeing that kind of experiment within the existing Fedora project; this is something that would be enabled by the Fedora.next proposal I'm making at Flock.
 
+Jef Spaleta Sub-community's have full editorial control over their spins.

They themselves decide how fast or slow they want to move and what goes in and what goes out based on whatever conscious they reach within themselves regardless how the overall community feels about it and regardless what FESCO decides.

It's up to them to decide if they deem a feature ready enough or not ready enough, to be included in their product for their target audience.

They are the ones doing the work.
They are the ones that are working at towards their goal targeted at their userbase through the product they work on presenting to the world.

Yes that is that simple...
 
Jóhann,  the reality of the level of creative control for spins given to spin maintainers doesn't change anything Lennart wrote, nor does it change the frustration level being expressed specifically at the FESCO decision making process.   Clearly Lennart is frustrated with the FESCO governance and featuring process specifically, even though the desktop spin can incorporate the changes he's pushing without any fuss and most likely will.

When frustration with elected governance is expressed, I don't think its constructive to reach for subversive tactics to undermine that governance...to deliberately undermine what is meant to be an accountable governance body, accountable through the elections process.

If there is a systemic difference of opinion that can be labelled as factions as Lennart has in his post... then I would prefer each faction work towards taking ownership of the rudder of governance by putting forward a coherent set of electable candidates and take ownership of the issues they care about. Present me as a voting member with a real slate of candidates and a real vision for where the feature process is taking us over the next couple of releases.    Don't just throw your hands up in anger at how the decionmaking process isn't working when it fails to work for you.  Own the process,and be accountable for it by standing up your vision so I can vote for it and help give you faction the mandate to make Fedora competitive.  if you don't want to be accountable to the body of the contributing membership, then don't. But please let's not deliberately subvert what little accountability we have built into what little governance we rely on.
 
+Lennart Poettering I totally understand your disappointment. Slowly but steadily I see the same thing happening with openSUSE, my favourite distro. Isn't it time for an RPM-based, desktop-targeted, modern Linux distro?
 
+Matthew Miller you know, the reason why I find the decision FESCO so uninteresting and why I am not willing to implement it is that it buys us nothing. *A change of defaults in something that is not the default is as good as changing nothing. *

People can call me childish as much as they want, but ultimately just shuffling things around in comps is deeply irrelevant, the goal of the feature I initially put together was to change the defaults, as that is what has an impact.

Of course, for you the cloud image is in the centre of interest, and that doesn't use the defaults anyway, so for you this is a welcome change, you got exactly what you want. But Fedora shouldn't just focus on that, but always keep the big picture in mind.
 
+Lennart Poettering I'm working on the cloud image, but you're mistaken in assuming that that's all that I want. I care about improving the distribution as a whole, and am willing to work to get there. I understand your frustration with the mailing list conversation; I guess my main comment is that that wasn't really necessary -- it's fine to make the proposal, answer serious questions, and just ignore the people who just constantly repeat the same negativity over and over.
 
+Matthew Miller believe me, my disappointment about your vote on this is much bigger than the frustration I got from the mailing list flamefest...
 
Well, again, I'm sorry about that disappointment. My job is to listen to people, not just one side, and having done that I felt that the responsible thing to do was to abstain. I hope despite the disappointment you are still interested in working to make things better. 
 
+Lennart Poettering about that booting in four seconds distro, is there some place to read/learn about that system, and maybe download to install it?
 
I tried to remove sendmail from the default twice (first around F-12) and gave up after too many flame fests.
 
+Lennart Poettering, I think you'll find that "those folks who talk a lot and want to decide and never do anything" exist everywhere. They are not an open source problem. They are a human political problem. You will find them everywhere. They gravitate to committees of all shapes and sizes, and tend to cause the formation of new ones. (I should know: I'm one of them by instinct, due to extreme innate conservatism. But at least I know I'm one of them, and try to fight it.)

In some industries, they have won control, and the consequences can be pretty horrible. A couple of years ago my uncle failed, after five months of effort, to get a single-column non-unique index added to a production database at his workplace. This terrifyingly radical change needed unanimous signoff from three committees with a total of fifty-plus members, most nontechnical. Getting signoff from that many people on anything short of a cure for death is probably impossible.)
 
You should join Debian for a while. The pull towards stagnation is so huge Fedora will feel leading edge again!
 
I said Debian is worse, actually ;) it will take so long to switch to systemd by default it's going to be comic.
Add a comment...