I hesitate to resurrect this thread. But recently this issue has tied in
to some things I have been learning about functional programming. The
description of systemd provided on the homepage is as follows:
[QUOTE]systemd is a system and service manager for Linux, compatible
with SysV and LSB init scripts. systemd provides aggressive
parallelization capabilities, uses socket and D-Bus activation for
starting services, offers on-demand starting of daemons, keeps track of
processes using Linux cgroups, supports snapshotting and restoring of
the system state, maintains mount and automount points and implements an
elaborate transactional dependency-based service control logic. It can
work as a drop-in replacement for sysvinit[/QUOTE]
In one school of thought, at least, the ideal programming solutions
should be highly modular, with code for distinct modules well isolated
from each other, and interaction (when needed or beneficial) between the
modules provided through straightforward external interfaces, rather
than allowing the innards of separate modules to become meshed and
tangled together.
The reason for this is primarily to avoid a very dangerous problem:
interactive complexity. That is to say, the more interaction there is
the between the inner components of the various modules (or groups of
functionality) the more potential there is for one module to affect
other modules in a way in which we did not anticipate. This complexity
grows in exponential proportion to the size of the code itself. In a
worse case scenario, a fundamental flaw in the design of one component
could leave the entire system crippled.
So the pressing question is: Does systemd conform ruthlessly to that
ever so critical modular approach? The simple fact that systemd provides
many diverse functions does not /necessarily/ mean that it is not
designed to be modular. For example, postfix actually provides a fairly
large range of functonality for one package. Nevertheless, postfix
achieves Unix-like modularity by literally dividing all core functions
into small, separate processes, with tightly defined interfaces.
Is systemd fully committed to this sacred approach? If so, it does not
advertise it well. Furthmore, it does not bode well when one reads
through each entry in the fledgling bug report list:
You will already find problems in interactions with components such as
the tty system, avahi, the mount system, and cryptsetup.
Anyway, I know others have said the same thing in different words, but I
hope that I have added something.
-- frigidcode.com theologia.indicium.us --------- To unsubscribe, send email to <aklug-request@aklug.org> with 'unsubscribe' in the message body.Received on Thu Apr 28 21:45:56 2011
This archive was generated by hypermail 2.1.8 : Thu Apr 28 2011 - 21:45:56 AKDT