DragonFly BSD
DragonFly kernel List (threaded) for 2004-01
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: what does you think of fink?


From: Pete Carah <pete@xxxxxxxxxxxx>
Date: Sun, 4 Jan 2004 20:45:28 -0800

To look at extremes...

If you want a good but *very* complicated system, I know of two that 
handled versioning just fine, and that I wouldn't want to reinvent:

SMPE on MVS (it took *all* your cpu power and several gig of ram (and a
3380 volume or two) a long time to sort out versioning on the entire 
system; I don't know if it would let you deal with only one package in a 
session or not.)  This was used to maintain *everything* - kernel ("nucleus") 
and everything else furnished by IBM and perhaps some other vendors.  
I'm not really recommending this but it's easily the most capable (and 
complicated) end-user versioning system I've ever heard of. 

INST on IRIX.  handled package versioning fairly well, and handled things 
like conflicting OS patches very well.  Of course the latter doesn't 
matter much when you do most maintainence from source the way we generally 
do.

Both of the above worked better than Solaris' patch+package system,
and infinitely better than FreeBSD's.

I don't know what VMS had for this purpose but I'd assume it was both
bloated and probably didn't do quite what anyone wanted...

Full dependency tracking is at least quadratic in the number of files in 
the system; maybe worse...  (this can even include user files like *.doc 
and such; they depend on both system modules and applications.)  Fortunately
one rarely needs to track *all* file interactions.

It actually could help to look at some of the history of this kind of 
thing; even if we don't want to repeat it there was some useful stuff
(and some awful stuff too) out there.

-- Pete



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