Re: More FUD for everyone: Computers Are Dangerous! (Users are devs, after all.)
On 23. sep. 2014 04:02, Joel Rees wrote:
[snipped]
Just to off-set the negative comments, I'd say Joel R. States his case
pretty well, though I think dropping the sarcasm-bit would have improved it.
I'll paraphrase my understanding of A FEW of Joel's points, just to see
if I read him correctly. Please let me know if I have misunderstood.
--------- A few of Joel's points: ---
1) Quite a few developers feel the need for training-wheels to get
inter-process-communication and process-control right.
2) Systemd is one way of gaining said training-wheels. The way that
seems to be the preferred among the majority,
2) Being monolithic, systemd ends up introducing inter-dependencies that
need not be introduced.
3) Putting in an extra layer to control settings in the kernel and
file-system introduces un-needed complexity and confusion.
--------
Now, for the sake of argument let's take as given that my understanding
of Joel is more-or-less correct, and furthermore that the points are valid.
** My thinking in response to theese points **
Given that systemd does have some negative side-effects, it would be
desirable to have an alternative. One of:
a) remove the need for training-wheels (raise the bar on entry to being
a developer, or give better training).
b) introduce an alternative set of training wheels.
c) stop debian development.
Even if neither a) or b) happens, nothing will be able to force c).
Nothing will ever force anything in an open source project. The only
"force" is the dynamic between number of developers willing to put in
the work, and number of users willing to use the result.
So, I guess this comes back to "show us the code" (or the howtos for
doing it differently).
Apparently Joel has (valid! ) other commitments. There are people
willing and able to work on systemd. Unless someone else steps up with
code or an online class in linux IPC, I'd say the case can be closed.
That "online class in linux IPC" would not only need to teach code that
works, it would need to capture mind-share to such an extent that most
people would say "that is the linux way". The situation to day, with
e.g. the /etc/postfix/master.cf as "state of the art", while competing
with grafted-on-crutches such as daemontools means a poor admin can not
stay within his comfort-zone if he wishes to run any box he may
encounter. Anything that reduces that kind of stress, will gain ground.
I think that any alternative, to succeed. would need to cover the same
broad base as systemd, either through code, or universally accepted
"best practice" on a number of areas. Starting a movement and getting
the man-power to do that is no easy task.
Reply to: