:: Re: [Dng] Packaging system
Góra strony
Delete this message
Reply to this message
Autor: T.J. Duchene
Data:  
Dla: dng
Temat: Re: [Dng] Packaging system
If I might add my two cents on the whole issue of Lennart, systemd and
package systems? While I do not believe in character assassination, I
do critique. Please do not bother reading if you want short answers or
polite remarks.


On Lennart:
Who cares what Lennart Poettering thinks? For that matter, who cares
what Linus Torvalds thinks? There are "buckets" of wasted electrons on
what those two think, and while I sometimes find them entertaining, the
whole "cult of personality" is a waste of time and energy.

The only time that someone's opinion should matter in open source is
when that opinion makes a material difference in what you are working
on. If that opinion does not work for you, fork the code and be done
with it. That is the whole point of open code. If anyone thinks that
these so called "luminaries" do not make bad judgements or crappy code,
think again. If the Linux kernel were perfect, it wouldn't have bugs.
The point of open code is that eventually the best code will hopefully
win out.

The problem with this particular form of Darwinism, and with Linux
culture in particular, is that everyone likes to talk about bad code and
only a tiny minority actually do anything about it. They would rather
coast on someone's bad work rather than create an alternative. That
lack of majority performing actual fixes is what is holding things up,
not someone's "vision".

On systemd:

On 12/22/2014 5:46 AM, Daniel Cegiełka wrote:
> btw. very good summary of systemd:
>
> http://openwall.com/lists/owl-users/2014/12/21/2
>

The idea of systemd is not a bad one, it simply suffers from a lack of 
good management.  A good project manager sets concrete goals and makes 
certain those goals are met.  That is what systemd is lacking.  That is 
why it is prone to mission creep and why it is repeating the mistakes of 
projects that have come before it.    I've never met either, so I can 
only judge based on what I can acknowledge, but I would say that in 
spite of his abrasive nature, Linus Torvalds is an excellent project 
manager, while someone like Lennart is not.


Systemd is really symptomatic of the Linux community, quite frankly.
Before anyone gets too upset, consider that Linux as an OS (not the
kernel it is named for) suffers from huge QA issues. My comments are
not a swipe at Linux, but an observation of what is a peculiarity with
the community as a whole. The Linux community accepts contributions
mostly based on enthusiasm. Learn C or Python, and have a Linux
install, and someone will probably take your code, regardless of overall
quality. I'm not saying that this is a bad thing, quite the reverse. A
lot of the time the code produced will be terrible, but it also takes
things in new directions, and ideas that would have never otherwise been
created. It is those gems among the dross that gives Linux its
advantage. If Linux distributions have any really severe detriment, it
is a lack of QA when they put it together.

On package systems:
Books could be written on the subject of packaging, so I'll just stick
with what we know, which would be Debian. As a clarification, Apt is a
dependency solver, and Dpkg is a package manager.

I've no particular fondness for either or the Debian package system,
since it is lacking in several areas. As far as Devuan goes, that is
what it has for the initial release, at least. I've already made
comment on how I believe Devuan should either fork or discard the Debian
packaging system, so I won't bother rehashing old remarks.

I don't think that Apt will ever depend on Systemd specifically. That
being said, I would not be surprised if some over-eagar Debian coder
creates future dependencies to better work with future versions of
Debian as long as Debian remains a systemd distribution. I think the
real question is: "Do we care?" I think that the answer to that should
be a resounding: "No."

I believe that within one or two releases, Devuan will be forced to
abandon any attempt to remain compatible with Debian, mostly because the
sheer weight of all the work-arounds that will have to be created as
systemd becomes more and more integrated into Debian with each future
release.