On 09/22/2014 05:07 AM, Jonathan Dowland wrote:
Hi Rob,

I saw the bug closed (via mail on -devel) and personally thought it shouldn't
have been. However when considering next steps my advice would be to leave bugs
alone for a short while and let things cool off.

It's important and useful to file bugs, but that in and of itself doesn't solve
problems - they just track problems. My advice, if you have spare time to spend
on trying to get this problem solved, would be to think beyond the bug filing,
and try to figure out which packages need to be modified in order to solve the
problem.

First you have to know what the design goals are. Without that, there can be no agreement on what packages have to be modified.

I don't think anything less than multi-init support would be worth pursuing at this point. There are a number of well-know arguments for and against it so I won't repeat them. This is just a list of steps I think it will take to accomplish it:

1) The goal is "modular Debian." Multi-init is the means to achieve it. Being tied to one init system is what caused Debian’s problems, and the replacement did not fix it. A modular system has to support all init systems, including systemd, clones and custom inits.

2) All init systems, to support multi-init, will probably have to support sysvinit as a backwards compatibility option, and systemd unit/service files or a translator for future compatibility.

3) All services obsoleted by systemd (like login) must be supported at least until they can be upgraded or replaced. They may have to support some systemd features for compatibility reasons.

4) To have the slightest chance of succeeding, I think modular Debian packages will have to go into unstable soon and be in the next testing cycle. Time is against it and it be too late already.

5) It would be a massive project, but the individual tasks are small and modular. I think it could succeed with sufficient buy-in. Even systemd proponents can see the value of modular Debian, and will accept the challenge of maintaining a "universal operating system."










--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/5420ca28.7010...@ix.netcom.com

Reply via email to