Linux: concerns over systemd adoption and Debian's decision to switch

Thu Oct 23 09:04:39 UTC 2014

On 2014-10-23 9:15, Matt Palmer wrote:
> On Wed, Oct 22, 2014 at 10:05:30PM -0500, Jeffrey Ollie wrote:
>> To achieve the level of integration that timedated has with the rest
>> of systemd would require more than just putting code into timedatectl
>> to write out /etc/ntpd.conf and starting a service.  timedated talks
>> to networkd (that
>> DHCP server that everyone is hating on as well) in real-time to
>> determine the state of the network and to get any NTP servers that
>> were sent in DHCP packets.  To do that for chronyd or ntpd in the same
>> way would require code changes and the systemd developers didn't want
>> to do the work,
> 
> This is the core problem with systemd, in my mind -- and what has gotten
> Linus, amongst other people, so thoroughly cheesed off with the systemd
> devs.  They don't play well with other children.  They don't appear
> particularly interested in reusing any existing code, because it's a lot
> more fun to write new code.  I'm a strong proponent of Joel Spolsky's views
> on rewrites (sorry, no URL, I'm on the train) and I don't doubt that all the

http://www.joelonsoftware.com/articles/fog0000000069.html

> same problems will come to haunt systemd on its way from being the new kid
> on the block to being legacy code[1].
> 
> - Matt
> 
> [1] A computer industry term which means, "it works".
> 
>