[mdlug] Systemd Should Be A Fork -- Why Is It Not?

[Adam Tauno Williams]

On Tue, 2015-09-01 at 09:21 -0400, A. Zimmer wrote:
> have been conscientious enough to not disturb the parent project
> with their excursions.  Forking provides potential innovation and 
> change without needlessly unsettling or burying established methods. 

You want to Integrate services [which is what Systemd does] without
'disturbing' them?   Code paths running off down forks and lack of a
cohesive architecture is what got us to the point of needing a Systemd.

And so you create a fork... if it is soon abandoned due to no
contributors what have you accomplished?

> The whims of a tiny clique of programmers, with the financial backing
> of selfish corporate interests, have been forcibly ensconced into an 
> otherwise free and open software environment.

By which you mean the primary contributors to the core LINUX platform?

> Why has there been no outcry?  

Because systemd solved a very real problem that has been nagging LINUX
systems for a decade, and getting progressively worse as servers [for a
long time very static things] become increasingly dynamic and adaptive.

> Why has there been no resistance to this odious imposition?

See above.  It is a solution not an imposition; the odious part is Sys
-Admins need the take the time to refresh their skills [aka - read the
documentation]

journalctl -xb vs. tail -f /var/log/messages ... this is what the fight
is really about.  And, as a system admin with 20+ years experience,
syslog is ***CRAP***; it has been an unsolved problem forever. Now we
have a journal - the powers-that-be be praised! Multi-line log entries,
actual audit security, signed messages, easy log management and
expiration, ... simply awesome.  And l-o-n-g overdue.  To make that
happen **somebody** needs to say THIS-IS-HOW-IT-WORKS; thankfully we
got that from experienced kernel developers and not some corporate
division interested in selling a tool.

> Let there be no mistake.  Systemd is yet a work in progress and its 
> goals are fundamental and far reaching.  

Yes, much like the LINUX kernel.  Which would 'never' have support for
multiple process machines or non-unified memory array architectures.

> Linux ecosystem will need to crucially depend on its monstrous and
> bloated service infrastructure.  

You mean the thin shim that helps applications and services use the
kernel's IPC architecture?

> I do not oppose systemd on either technical or philosophical grounds.
> I disfavor systemd mainly because it diminishes the freedom of
> choice.
> My current Linux systems do not use systemd and it is my intention to
> keep things that way forever.

All my applications and services seems to run on Systemd hosts without
issue.  And that is the purpose of an Operating System.

-- 
Adam Tauno Williams <mailto:awilliam at whitemice.org> GPG D95ED383
Systems Administrator, Python Developer, LPI / NCLA