Re: setjmp/lonjmp
Michel Talon wrote:
> There are good reasons to remove things "that work" as you are saying.
> First, if you have 3 ways to do the same thing, this gives 3 times more work
> to the developers to maintain the stuff in view of the evolutions elsewhere
> (which often requires patching each of the 3 programs).
You aren't talking about real world open-source.
In real world, if there are 3 motivated and skilled potential project
leaders, exactly 3 "ways to do the same thing" are likely to turn out.
These, hopefully, will be good AND different products, possibly
incorporating popular features from each other over the time.
In _managed_ environment, this isn't big problem. But who've seen real
managed environment in open-source? (in apache perhaps?) There are ways
to make things not happen but (virtually) no ways to make things happen
(as planned).
And I beg to disagree over the "same thing" business. IPFW isn't the
same as IPFILTER, nor it is the same as PF. If they were, there wouldn't
exist three of them, for several years already.
> Second it introduces confusion for the users who have to choose between
> the 3 programs, without knowing which is the best, while they would be
> much happier that the developers choose for them.
So now it's users we are concerned about?
All right! Be concerned about ME! :)
I'm (potential) user. I've been already using ipfw (after choosing it
over ipfilter) for some years now. I'm studying the possibilities of dfbsd.
> Third it gives the impression of un unmaintained and crappy codebase, and
> this is bad. Let us look at the 3 firewall packages in FreeBSD-5.3. Only
Let me tell you -- from the user point of view -- that user doesn't
usually care a bit about such kinds of things.
What (open-source) user wants to know usually is:
- "is this piece of technology worth my attention (and climbing the
usually steep learning-curve)?"
- "is this piece of technology going to stay there and be maintained so
I don't waste my time learning it?"
THAT'S when things can start looking bad and damage reputations.
> one of them has been fine grained locked, i.e. pf. At the same time pf
> is coupled with altq which is notoriously the best traffic shaping utility
> available in FreeBSD. Conclusion, ipfilter and ipfw2 are clearly
And altq was awarded the title by whom? :)
> deprecated. Of course this is none of my business, but i suspect that
And this was decided by whom? :)
> Dragonflybsd developers who are less numerous than Freebsd ones,
> are even less inclined to maintain stuff which appears of secondary
> interest.
OH.
--regards
討論串 (同標題文章)
完整討論串 (本文為第 8 之 12 篇)：