[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Another system management tool to disappear.



On Tue, Sep 1, 2015 at 5:03 AM, T.J. Duchene <t.j.duchene@gmail.com> wrote:
>
> On 08/31/2015 05:14 AM, Joel Rees wrote:
>>
>> Actually, there's a couple or three questions going begging here, that
>> I'd like to ask:
>
> Sure, ask away! =)
>
>>
>> (1) TJ, have you ever built LFS? Or, even better, built a running OS
>> on top of the Linux kernel without even the help of the LFS tutorial
>> and tool set?
>
> No, I have never used LFS.

Yeah.

> I have, however rebuilt or otherwise modified:
> Debian, Gentoo, RedHat and others over the last couple decades.

There is a difference between what I asked and what you're telling me.

Simply tweaking and recompiling debian or redhat is not what I'm
asking about, although it can be tedious enough. Nor is building a
functioning gentoo really.

I'm asking if you have built an OS from scratch, including the
userland tools and apps, for a specific, non-trivial purpose.

I'm not asking for your CV/resume.

>  That is
> not including other things Unix: like Solaris. No, I have not always had
> documentation and sometimes had to figure it would myself.

Documentation is not really the issue.

> Is there a more specific answer you wanted?

You have already given me your answer.

>> (1a) If you have, have you ever implemented your own init system for a
>> Linux-based OS that you built yourself?
>
> No, I never had a reason to.

Clearly.

> As with many things, necessity breeds invention.  I have had no reason to
> invent my own when I can modify an existing one to do what I want.

And there we have, in a nutshell, why it's a little disingenuous of
you to raise the "You can always build your own!" argument.

You haven't done this one.

> With
> respect,

Should I believe you when you say that? (I know it seems to be picky
of me, but I've often found that this particular expression is used
more in the ironic mood. So I ask. Not that it's fair of me to ask,
because I know it's not a question that can be answered meaningfully.
But please don't ask me to assume that assertion means anything,
either.)

> I doubt most programmers would bother creating an entirely new init
> unless they had a pressing need or just wanted something new.  The whole
> point of open source is adaptation.

Perhaps it is to you. But if I needed only adaptation, the Macintosh
is a much more comfortable environment to do the adaptation thing in.

I have other needs. Unfortunately, there is no current OS/community
that can provide me those needs. The nice, though uncomfortable, thing
about the systemd business is that it brought my attention to that
fact.

> There are quite a few inits to chose from.  The fact that Systemd was
> created in addition to the dozen or so previously existed probably had more
> to do with cgroups than anything else if you ask me.

Well, I never said I cared much for cgroups, either. Quite the opposite, really.

cgroups is, in fact, part of the stuff I specifically do not need in my OS.

>> (2) Having done that much, have you ever kept that system maintained
>> and updated, even at just the level of keeping only the critical
>> applications patched or updated against vulnerabilities on a timely
>> basis?
>
> Yes, I have.

Well, ...

>  I used to manage servers for ISPs.  Yes, I'd even patched them
> by hand because the OEM no longer provided updates.

Hey, we've all managed servers and/or workstations here, I think. Or
we are learning how. That's not the question I'm asking.

And, since you haven't built the OS from scratch, ...

No. I beg to disagree with you, but I don't think you have maintained
an OS you've built from scratch. Sorry.

Your CV looks promising, but that's not what I'm asking you about.

>> Okay, there's actually one more question here:
>>
>> (3) Have you ever done the first two while holding down a full-time,
>> 40+ hour a week job that doesn't particularly make allowances for
>> employees that need to spend the time necessary for maintaining their
>> OS?
>
> Well, I can honestly say "No."  As I said, I have never bothered to write a
> new init from scratch.

There it is.

> What you are really asking

Please don't put words in my mouth.

> is when I was working other jobs as we all have,
> and maintain my own systems as best I could on my own time.  Sure.  We all
> do the best we can.  None of us are perfect and I have never claimed to be
> either.

Perfection is hard to achieve, as el viejo used to say. It's also not
really what I'm asking about.

>> If you have, how long did you keep it up without developing
>> personality issues for lack of sleep, developing dysfunctional
>> digestion problems like ulcers and diabetes, and/or ending up breaking
>> up your family?
>
> Well, to be perfectly honest, I do have some of those problems. Some are bad
> enough to where I am probably on medication for the rest of my life.

Sorry to hear that. There are doctors who want to get me on
medications for life, too. Fortunately, I know just enough medicine to
avoid needing what they sell.

I don't think they are particularly evil, but what they want me to
take would kill me. Not immediately, just by gradually making it
impossible for me to keep any sort of job at all. Maybe my
non-standard health has something to do with my non-standard
requirements in my OS. You know, personality does impact technology.

> I even
> have a few others heaped on top of them that you didn't mention, like
> cerebral palsy and arthritis. Actually, cerebral palsy is why I got into
> computers in the first place.
>
> I understand why you asked.

Again, don't put words in my mouth. It makes it very difficult to communicate.

> You are probably wondering if I have
> unreasonable expectations of others.  I don't.  I don't expect someone like
> Doug to compile everything from scratch,

Now, you see, you and I have a very different perspective on things
here. I don't want to even put Doug into the position of having to
wonder whether he should learn how to build his own OS from scratch.
At least, not now, when it would not be very meaningful to do so.

That's why, when asked, I've always advocated that the
systemd/cgroups/udev/etc. crowd be kind enough to set up their own
separate distros, instead of insisting that their thing must be
included as a by-default part of all major distributions.

> or you to rebuild Debian by
> yourself.

I hope you realize that you've kind of misjudged something here.

> Conversely, I do expect anyone - myself included - to back up
> what they are saying with good reasons and at least some experience.

So, you'd like to hear my CV?

Sorry, it's not exciting. It would be a bit difficult for you to judge
my skill level by it anyway. You can look up some of my back-burner
projects on sourceforge.org and osdn.jp. Not that it would tell you
much.

> I also prefer that in discussions of this nature, that people maintain some
> logical distance - separating the person from the code.

No self-respecting engineer will claim that the code he produces can
be separated from the context he operates in, including his
personality.

Sorry if I'm being too blunt with that.

> Lennart Poettering
> is not systemd and systemd is not Lennart Poettering. If that is not
> possible, then I really see no point in continuing.

Which, I am afraid, puts you in the pattern of those who defend the
approach the cabal has taken.

(Concerning the use of "cabal":

http://0pointer.net/blog/revisiting-how-we-put-together-linux-systems.html

Although I'm not really sure they understood what their own choice of
words seemed to imply.)

I will agree that some people have made too much of personality
issues. Theo is not a perfect person, but he's the right guy to lead
openbsd. Steve Jobs was not a perfect person, but he was the right guy
to lead Apple. The cabal is right for systemd, et. al.

The issue is not whether they should be leading their projects or not,
the issue is whether what they build is really appropriate for
becoming a necessary part of all major distributions of Linux kernel
OSses.

> A lot of people posting over systemd forget or do not realize a lot of
> important details - for example:
>
> 1. Many other people had added and subtracted from the code before you or I
> ever got our hands on it, including Debian.

What relevance does that have? Does it somehow cease to be systemd
when people outside the project compile it?

> 2. GCC is also a finicky beast, and is hardly bug free.  It  matters what
> compiler is used, even when compiling the kernel, much less anything else.
> You can introduce bugs into software just by using a different version of
> GCC than what the developers are using.

What relevance does that have? (Other than that, if I find gcc version
changes causing errors in my code, I go looking for what I messed up.
Without fail, I've found real bugs in my source code.)

> 3. The management tools for systemd are written in Python.  I personally
> find it a very questionable choice.  It can be considered famous for hard to
> find runtime bugs.

Python seems to me a very natural match for systemd, and the
personalities of those who develop it.

That's actually a good thing more than a bad thing. People should use
tools they understand.

> Take care!
> T.J.

You, too.

-- 
Joel Rees

You know, I really can't say reinventing wheels is evil.
http://reiisi.blogspot.jp/2015/08/why-i-cant-use-phrase-re-inventing.html


Reply to: