Re: proper place to discuss kernel 'bloatedness'?

Alexander Viro (viro@math.psu.edu)
Sun, 31 Jan 1999 20:30:08 -0500 (EST)


On Mon, 1 Feb 1999, Marek Habersack wrote:

> Well, in theory, it could be possible to construct a kind of a system to
> provide "partial" downloads for "partial" compiles. I imagine that one would
> download a "config pack" from the net, do make config wich would contain,
> basically, dependencies of all the files from the kernel sources needed to
> generate a custom kernel for the submitting user. Such file would be sent to a
> server (of course one such server wouldn't suffice) and the user would
> retrieve only a required subset of the kernel source tree, compile the kernel
> and be happy. It would be further improved by using rdist or similar
> mechanism, assuming that the user won't change his machine's configuration too
> often, the next source downloads would be as easy as retrieving patches to the
> changed files.
> It looks nice in theory, but would probably require creating a protocol to
> serve the config submissions and retrieval of the required files.

If you are OK with local compiles it can be trivially done via
CVS and skeleton Makefiles (think of *BSD ports). Since we have recursive
make anyway we could do download on demand from make. Actually I think
that it would be useful for normal setup too - s/download/ln(1)/. I'll
try to look at it - make definitely spends way too many time checking
dependencies. Smaller tree would make compiles faster.
Server-side compiles... ;-/ Nah... gcc is waay too memory-thirsty
and several clients will suffocate any server. Unless you have a spare
Beowulf cluster to burn, that is.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/