Archives and Packages (was Re: Universal and binary builds)

[Jordan K. Hubbard]

On Mar 28, 2009, at 5:10 PM, Jeff Johnson wrote:

> I'm saying that you were largely honking the same Newer! Better!  
> Bestest!
> themes 5+ years ago.

Hurm.  If that was your take-away from this, then I somehow  
*seriously* failed to make my point, both 5+ years ago and now.  I am  
not asking, nor have I ever asked, for "Newer!  Better!   
Bestest!" (assuming, of course, that I even know what you meant by  
that ;-).  I'm asking for *minimum requirements* for mainstream  
success and that's all I'm asking for.

Now, perhaps MacPorts does not want to go mainstream, much less  
achieve mainstream success.  I've voiced that particular theory more  
than once myself, and don't forget: I've watched MacPorts go from an  
Apple internal project, done by some fraction of an FTE spread across  
the 2-3 of us who worked on it sporadically on and off the clock, to  
an external, purely volunteer-driven project which looked for quite  
some time like it was simply going to hit the ground with a meaty thud  
and DIE, to a purely volunteer-driven project with some actual  
volunteers and a rather amazing resurrection, as Open Source projects  
go.  Once OSS projects get seriously ill, they generally just die,  
they don't recover and go onto even greater success than they had at  
initial launch, so sure.  I could easily see the MacPorts project  
saying, in some collective consciousness fashion: "Hey, we nearly died  
and came back to life!  We have a lot more ports than we ever did  
before, a lot of them even work now, so hey, what the hell do you  
want, BLOOD?  Go peddle your binary packages somewhere else!  We're  
busy!"   I can see a whole lot of justification for that point of  
view, which is why it's always with a sense of unease and mixed  
feelings that I even get into this whole, stupid packaging discussion  
from time to time. :-)

That said, should MacPorts ever DO decide to go from having thousands  
of users to having hundreds of thousands, or even millions, of users,  
I don't think I'm way out of line in suggesting that one reason will  
be because checklist items 1-4 were finally checked off by somebody.   
It CAN be done, Anders' tales of woe from previous MacPorts attempts  
in this area notwithstanding, with projects like Debian proving it  
every day (due to the innate superiority of their packaging tools, I  
suspect).

OK, that bit in parenthesis was just to piss Jeff off, I didn't really  
mean it. ;-)

> And MacPorts still has none of the infrastructure that you've  
> outlined,
> however reasonable your goals may be. The goals, in fact, are  
> reasonable.

...  and we could work backwards from this statement in search of any  
number of potential causes ("they don't care!", "they don't know  
how!", "they didn't use RPM 6!") but I'm not sure it would get us  
anywhere. :-)   I suspect this is simply one of those "when its time  
comes, it will happen, and if its time *doesn't* come then, well, it  
won't" sorts of things.

> And truly, my only quibble is with "completed", there's certainly huge
> differences between rpm and xpkg no matter how much you wish to
> lump them together.

Well, seeing as how xpkg does not, in fact, exist yet, except as ideas  
in various peoples heads, I'm not quite sure how you can compare it  
with *any* other system.  They are only in common association through  
the term "package manager", and I don't think that anyone, least of  
all myself, was attempting to compare, say, xpkg's built-in flux- 
capacitor based time-travel features with RPM 6's adoption of Perl6 as  
an AI language which simply guesses what the user wants before they  
even install the package, thus pre-installing it and yielding the same  
results in a very different way.

- Jordan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macosforge.org/pipermail/macports-dev/attachments/20090328/b358f052/attachment-0001.html>