[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: A few observations about systemd

Bernhard R. Link <brlink <at> debian.org> writes:
* Uoti Urpala <uoti.urpala@pp1.inet.fi> [110720 18:37]:
>> Supporting things like kFreeBSD is a lot of effort to benefit few people. If
>> it's what a volunteer wants to work on then that is his right. But to insist
>> that others should work on it is wrong; project-wide priorities should be
>> based on rational decisions instead.

> Every single one of us and our users is part of some set of "few people".
> It is not "rational" to say "this only benefits a few" because if you
> only do things each benefiting only a majority, the intersection of all
> those majorities will be very small. And the amount of skilled
> developers within that set most likely even smaller.

There are two primary mistakes you're making here.
First, as I already explained in earlier mails, it's wrong to think of only very
visible features like a separate kernel as counting for excluding/including the
people affected. There are _hundreds_ of other possible features/fixes with
corresponding groups of people as large as potential kFreeBSD users. It's
impossible to support them all. You need to prioritize. To point to kFreeBSD and
say "this must be supported because not supporting THIS would exclude people"
(as if that wouldn't be the case for all the hundreds of other things) is wrong.
Second, you're ignoring the negative effects of the decision to support
something. Saying "the project must support this" does not magically add
features that help someone without hurting anyone. To make a rational decision
about kFreeBSD support I can consider the likelihood of needing kFreeBSD in the
future vs the likelihood of encountering some problem on Linux that would have
been fixed had resources not been diverted to kFreeBSD support, or would never
have appeared in the first place if not for compromises needed for the sake of
kFreeBSD support. The latter probability is much higher.

To put it in software development terms: you are advocating feature creep.
Although it can appear otherwise at first, feature creep does not result in
programs that are useful for a larger amount of people. In the end the result is
less useful for almost everyone.

> And for the rhetoric: It's not about insisting people should do other
> people's work, but about people should not work against other people's
> work.

It's not that anyone would break kFreeBSD for the sake of breaking kFreeBSD.
Rather the most natural and efficient way to implement systemd support on Linux
leads to kFreeBSD breaking as a side effect. The existence of kFreeBSD should
not block this work or make it harder.

> We need to all together create a project we all can live in. Just
> playing the "I do not care about that, that does not benefit me and the
> users I care about" card does not create a project but only poisons
> the community.

It's impossible to support everything. And supporting kFreeBSD _does_ hurt
others. It's possible to support both kFreeBSD and systemd, but only at a cost
which means something more important than kFreeBSD will suffer. If someone
volunteers to work specifically on that, fine; but it would be irrational for
the project to favor that.

Reply to: