Openness vs. Comfort

Fri Jun 12 15:45:06 UTC 2020

On 6/12/20 8:37 AM, Vincent DEFERT wrote:
> Hi,
> 
> The "freebsd vs. netbsd" thread and its flame war about top posting made 
> me think seriously about what I've perceived in the FreeBSD community 
> over the last 9 months.
> 
> My impression is that there are 2 sub-groups in the FreeBSD community, 
> those wanting FreeBSD to dominate over Windows and Linux, and those 
> wanting to keep every semi-colon in its set line and column forever.
> 
> [ Deliberate exaggeration here, but posts on this mailing-list and on 
> the forums seldom fall in the mid-range. ]
> 
> I tend to think that this schism is caused by a lack of common vision. 
> And without a vision, you have no criteria to make decisions, so you end 
> up battling over and over with little results.
> 
> With a vision, it would be easy to decide what is important and what is 
> not, but there are a couple of obvious things that can be discussed 
> already.
> 
> 1. Why mailing lists?
> 
> I assume all of you have perfectly healthy eyes. Great!
> Unfortunately, this is not my case. For me, reading plain text messages 
> is a torture. I made an effort in the beginning, but it is not possible 
> in the long term.
> However, as healthy as you may be, I've read complaints about the person 
> recently posting from his smartphone - and yes, his messages were in 
> effect as unreadable as the replies.
> 
> But what purpose do these mailing lists serve?
> I've read several times they were the right place to meet developers, 
> but this is mostly true for freebsd-current@ an freebsd-stable@, not 
> this one.
> 
> So in the end, this mailing list (freebsd-questions@) is just to be used 
> when looking for help.
> Which means that at the time you need help, you must face additional 
> difficulties, at least legibility and focus.
> So one should only use this list when not in need of high importance help.
> 
> I don't know what you do when you have to solve a problem, but the first 
> thing most people do is google for it.
> And what do you get when you google something? Links to web pages.
> Which mean that if answers given on this list had some value, they will 
> be invisible to most people.
> 
> So using a mailing list in 2020 is reserved to things of little or no 
> value.
> Important things deserve a forum: they are visible (indexed by search 
> engines), legible (web browsers accessibility features) and well 
> structured (threads) so you can stay focused on what you're here for.
> 
> 2. Linuxophobia / Linuxallergia

Disagree. I did flee Linux, but only for servers. Desktop/laptops and 
number crunchers I support are still Linux.

> 
> If I were fully satisfied with Linux, I wouldn't be there.
> However, there are also good things in the Linux world that could 
> inspire development decisions for FreeBSD.
> But many people in the FreeBSD community seem to suffer from severe 
> Linuxophobia and/or Linuxallergia,

Huge turn made by Linux when systemd and friends were pushed down our 
throats will excuse the above feelings of those who do have them.

> throwing out the baby with the bath 
> water.
> So what do we get in 2020?
> A ports collection with a huge dependency mess and unreliable package 
> repositories that remove your applications when a build has failed.

Disagree. Even though port collection shouldn't be confused with FreeBSD 
system, it still is great in my book. I maintain FreeBSD servers for 
multiple years, and services are based on potrs originally, and on 
packages now; and those ports that do not have resembling packages 
(there usually is reason for that), and those which I need built with 
different options compared to what pkg is built with I build with poudriere.

I did not observe grave failures here, occasionally you need to do a bit 
of work and use a bit of brain, and that has been a routine thing when 
ours was almost exclusively Linux shop, and is now when most important 
stuff runs under FreeBSD,

> This could be admissible in the 90s, but not in 2020.
> 
> When you report these issues, you're told "jail everything" or "use 
> poudriere".

It is a fact of life that if you attempt to make big enough collection 
of things installed, you imminently bring dependencies that will 
contradict each other. E.g., by installing binary or library with the 
same name into the same place. Now, can anything be done about that, and 
will it be reasonable to attempt?

Let's first compare with excellent Linix distro: RedHat Enterprise Linux 
(RHEL), or its "binary clone" CentOS. What you can install from main rpm 
repository virtually never has problems. It is possible, because main 
repository has really small number of packages - I didn't count, but I 
bet ports tree has an order of magnitude larger count. And to acieve 
that RedHat puts into it a lot of work, even though the collection is 
not large.

Is it reasonable to exect the same from FreeBSD ports collection? No, it 
is not in my opinion.

First, logistically, ports are independent, maintained usually by 
different maintainers, and who is to arbitrate which of the ports should 
change what it produces in case of conflict. And with ports come and go 
(you might notice ports deprecation) how it can be judged which port is 
to stay for long, and hence is more important if arbitration is done?

All in all, ports (and packages) as they are are excellent means to 
build what you want your machine to be. With FreeBSD you are much more 
flexible. But to enjoy that one has to accept some need to do work and 
thinking.

If one aims at "no effort" installation, then "turn key" products are 
more reasonable for consumption. Such can be MacOS, RedHat Enterprise 
Linux, and varioety of "signe task" boxes such as wireless routers, 
"hardware" firewall boxes etc. They all need little to no effort (and 
some are based on the same FreeBSD, other may be on OpenBSD or Linux). 
But they have really restricted function, the one that one was put 
together for.

> Those who do so set strong barriers around FreeBSD.
> 
> Using jails means every new user must learn a whole lot of things to use 
> FreeBSD, even in irrelevant use cases.
> Using poudriere means learning even more + dedicating a machine to build 
> your packages + waiting for as long as needed to build everything you 
> need + fixing bugs + rebuilding.

New user can begin without that.

> 
> It means FreeBSD imposes on its new users a tremendous cost - a cost 
> which is incurred only ONCE for each Linux distribution for the benefit 
> of all its users.
> Linux distributions have fully addressed these issues 15 years ago and 
> it is the bare minimum expected from a distribution.
> At least for this, Linux would be a good source of inspiration. And I 
> know at least one Linux package management system released under the BSD 
> license.

With thins big statement one will think that it will be counter 
productive to give you different arguments. Sound like you have arrived 
at your decision. It is Linux for you.

With that in mind the why this your post? With the goal to start new 
flame war? Should I have guessed that from that nice subject?

> 
> The evolution of the IT landscape over the last decade shows a dramatic 
> loss of appetite in more and more people for reinventing the wheel over 
> and over.
> 
> 3. Comfort and Openness
> 
> FreeBSD has a great base system and a great text mode installer, but 
> what's the point in installing it if managing applications is a mess and 
> asking for help a curse?

Making statement "is a mess" is objectional. It is not for majority of 
people, and as it is mess for ytou, it must be something about you, not 
about system. I'm starting to think that I have already talked to you on 
this list, only you were behind different email address then.

> Obviously, what is privileged in the making of development decisions is 
> the comfort of first time FreeBSD adopters.
> However, comfort is a great thing as long as it doesn't turn into rigidity.
> Conversely, life requires from a species a minimal interest in its 
> environment and a minimal ability to adapt.
> There is a word for this: openness.
> 
> Being open (or opening up) doesn't mean giving up on what matters to you.
> It just means you know quite well what matters to you and you feel safe 
> considering what surrounds you, and use whatever out there you deem 
> appropriate to take good and continued care of what matters to you.
> 
> 4. And so what?
> 
> Nothing.
> My sole purpose was to provide you with an insight of how FreeBSD and 
> its community could be perceived by an outsider in 2020.
> I'm pretty sure it is of interest to some of you, it's the only reason I 
> wrote this mail.
> But in the end, interested or not, what you do or don't do with this 
> piece of information is yours, not mine.

That is nice "detached" attitude. Indeed, BSD descendants can be 
percieved like that by some people. And the same as such people exist, 
there exist people for whom Linux is perceived quite similarly. That is 
not a big secret: in large enough number of people one can find the 
whole spectrum of opinions abut any given thing. There is one thing 
though: some opinions have no grounds at all. Some may have some grouds 
as person tried something, but didn't find what expected, like your case 
probably. Some do have grouds after long use and constant putting effort 
into making things work to one's needs, and realizing how much one may 
still need to learn.

====

I hope, this your post (neither my replies) starts new flame war on the 
list. We really have better things to do. But alas, FreeBSD lists do not 
even require to sign up to post (now tell me about openness!), so anyone 
can start anything on FreeBSD lists.

====

Valeri

> 
> Vincent
> 
> PS: These topics are not FreeBSD-specific, they apply to all BSD OS, the 
> situation of the others is just much more degraded.
> _______________________________________________
> freebsd-questions at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to 
> "freebsd-questions-unsubscribe at freebsd.org"

-- 
++++++++++++++++++++++++++++++++++++++++
Valeri Galtsev
Sr System Administrator
Department of Astronomy and Astrophysics
Kavli Institute for Cosmological Physics
University of Chicago
Phone: 773-702-4247
++++++++++++++++++++++++++++++++++++++++