busybox Reboot

[Rob Landley]

On Tuesday 28 February 2006 3:34 am, Andy Green wrote:
> Rob Landley wrote:
> > On Monday 27 February 2006 1:06 pm, Andy Green wrote:
> >> Ole-Egil Hvitmyren wrote:
> >>> There is one problem with that.
> >>>
> >>> A WHOLE lot of stuff is COMPLETELY rewritten (functionality wise, not
> >>
> >> I don't think you could sleep nights anyway knowing that you are relying
> >> on such old and deprecated versions and not really therefore in control
> >> of your code.
> >
> > Could you not insult our userbase, please?
>
> This wasn't intended as an insult.  You can either track bitrot piece by
> piece or take a huge hit (which this guy therefore tries to avoid)
> bowing to the inevitable.  If you don't keep up with the projects being
> used then you are on a "Legacy Loser" sooner or later.

This is why I like gigantic regression test suites.

And the "bitrot" you refer to only happens when you upgrade only some pieces.  
If you upgrade everything, it should work (and if it doesn't, tell us and 
we'll fix it).  If you upgrade nothing, it should work.  If you upgrade some 
things and not others, that can break and we may not be very sympathetic.

> > Also, stop and think about what you just said.  The majority of the
> > planet is currently using Windows.  (The most recent major release of
> > which, XP, was in 2001.)  If people can sleep at night running that stuff
> > (perhaps with the help of alcohol), a five year old version of busybox
> > looks pretty nice in comparison.
>
> Most of those users are not shipping infrastructure products with
> Windows XP in them though; examine what this guy's company actually
> does.  The nature of the guy's problem - random reboots of his box -
> suggests to me he has indeed lost control.

It could also suggest that he has an accumulation of dust in his router.

> I get antsy when code I am 
> relying on has had a few revisions that I did not track because I know
> it means that my project breaks if it leaves its little fly-in-amber
> time cell, or if I have to go back to it then I face trouble if I try to
> fix problems by taking later code: my advice to this guy reflects that.

That would be the "upgrade some things".

If you have the source, you can backport individual patches to old stuff 
yourself.  That's what Red Hat used to do, and Fedora Legacy exists for this, 
and so on.

> > He has more control using busybox since he has the complete source source
>
> Sure, having source gives you that opportunity, if you can take it.
>
> >>> If I cannot get any support without upgrading, then I might have to
> >>> say no to using your product
> >>
> >> Steady on, some of these guys might lose their bonus!  Lol
> >
> > The fact you don't care doesn't mean the rest of us don't.  We actually
> > do support our old users, that's why we have the old versions online. 
> > (We may support them by advising them to upgrade or quoting consulting
> > rates for doing work that doesn't help the project as a whole.  But
> > they're not going to be abandoned.)
>
> Well good for you.  However what you parsed as "not caring" was in fact
> amusement at the guy deploying commercial-world logic on this GPL
> project so he could avoid, or, well, put off, coming up to date.  It
> sounds like you're open to that so I guess the joke was on me.

If he's willing to pay somebody to support it, then GPL software is indeed 
part of the commercial word.  If he's trying to do the half and half thing 
again (support for really old stuff, support for free), it's unlikely to 
work.

We support free software by fixing anything that's wrong with it.  If we 
already _did_ fix whatever was wrong with it, our response is generally "we 
fixed that already, upgrade please".

That said, we've been known to go above and beyond the call of duty just to be 
social, but don't base a business plan on that.

> -Andy

Rob
-- 
Never bet against the cheap plastic solution.