Re: testing wants to install systemd
On 11/20/13, Joel Rees <joel.rees@gmail.com> wrote:
> On Wed, Nov 20, 2013 at 8:24 PM, Zenaan Harkness <zen@freedbms.net> wrote:
>> On 11/20/13, Ralf Mardorf <ralf.mardorf@alice-dsl.net> wrote:
>>> On Wed, 2013-11-20 at 21:00 +1100, Zenaan Harkness wrote:
>>>> http://0pointer.de/blog/projects/the-biggest-myths.html
>>
>>> I'm using systemd for a very long time now, the content of the above
>>> link is complete bogus, since it does ignore the real issues.
>>>
>>> However, I won't discuss it again.
>>
>> If you post a link to these "real issues" you discussed, then that is
>> useful for a constructive discussion.
>>
>> Otherwise, your "real issues" are just hearsay/ handwaving.
>> [...]
>
> I guess, Zeenan, that you are trying to be reasonable, since you admit
> that you too had problems in the transition and end up not using it
> now. (I did understand you right about that?)
Yes, this is correct.
> But the biggest-myths link you posted is the systemd leader himself
> engaging in a litany of naked assertions, telling the world why his
> baby is not ugly. Naked assertions have no particular virtue over
> hearsay and other forms of handwaving.
I understand this. However the biggest-myths page is _technical_
assertions and 'hand-waving' if it must be called that. Ralph was
posting "emotional" hand-wave, with no technical content, which I say
is not useful to me to determine if I am not understanding something
deficient in systemd (which at the least, I personally faced at least
one deficiency too significant for me to continue use).
> The closest thing he offers to proof of any of his points is
> unsubstantiated boot-up times. Now, if the rest of his assertions were
> as commonly accepted as the boot-up times, we could overlook the
> general hubris in that post.
>
> One minor quibble with his myths, the *nix shell languages are not
> arcane, no more arcane, at any rate, than C itself. The odd syntax for
> conditionals has a reason. Every programming language has it's
> reasons, and failure to understand them makes them appear odd. But odd
> is not arcane.
I agree with this.
> Now, perl can be arcane, but Lennart didn't address perl at all. He
> didn't really address sh either, come to think of it, just waved his
> hands at it.
>
> Program source can be arcane in any language, and this is one point he
> totally misses. It's kind of representative of the way he keeps
> failing to see the forest for the trees --
OK. Thank you. There is a point I am starting to see. And I agree it
appears Lennart overlooks the 'complexity' of C code in comparison to
sh code, and implicitly (without realising his own assumption I think)
asks for forgiveness because "the unit files are so awesome and clean"
or something...
> Initialization files have syntax. They may not form turing complete
> languages, but they do form a language. (Yes, we call XML a language.)
> Any language can be used in arcane ways. The quickest way to make a
> language arcane is to try to force it to into contexts the language
> design ignored.
Very true.
> That's one of the things that ends up monolithic about systemd, by the
> way, forcing the common init syntax.
>From my reading, and limited experience, I must take exception with
"forcing" - by design, systemd does no more than "encourage" common
(and fully declarative) 'unit file' syntax, and I say (from my limited
experience) that it does this in a _very_ good way. Declarative and
consistent configuration (where it is 'sensible' to do so - ie not
shoe-horning procedural stuff into the declarative config, as you say
above), is a good thing to strive for - I fully support this goal of
systemd, and I have to say, it is not forced (as far as I can tell).
> I realize I am starting a deconstruction of his arguments, and I don't
> have time for that.
Don't worry about it. I appreciate that you bring something to the
table that is constructive. For me this is really appreciated so thank
you.
> Systemd offers a framework. If Lennart had not been so insistent that
> everyone had to test his baby now, on their production systems, we
I don't think he was the one who insisted, by I haven't read many
online discussions either. I think RedHat (and Fedora, and Arch, and
...) all have their own decision making processes, and although I'm
sure Lennart was involved in some of those, it is inconceivable that
he could "insist" in a forcing way, his will upon RHEL, Fedora, Arch,
etc etc. Inconceivable. I apologize if this is a strawman and not what
you said, but it sounded like that.
> might have had time to refine the framework as a community. As it is,
> I'm not as optimistic about it as Ralf.
:)
Joel, again, thank you. I really appreciate your insight - it brought
something to light I did not see before (yes, I felt it, but I was not
able to name it, perhaps now I can begin to name it).
Regards
Zenaan
Reply to: