Bootstrapping X.Org (was Call Monday 24 Jan 2005)

[Torrey Lyons]

At 9:01 PM -0500 1/25/05, Adam Jackson wrote:
>On Tuesday 25 January 2005 20:44, Torrey Lyons wrote:
>  > At 7:35 PM -0500 1/25/05, Adam Jackson wrote:
>  > >As a developer, I want the world to build as little as possible, because
>>  > at any given moment I care only about the one piece I'm working on.  If I
>>  > really want to build everything in the world:
>>  >
>>  >cd ${xapps}
>>  >for i in */
>>  >do
>>  >  pushd $i
>>  >  make
>>  >  sudo make install
>>  >  popd
>>  >done
>>  >
>>  >It's _trivial_.
>>
>>  You have some good arguments, but this is disingenuous. There are
>>  dependencies between the parts of xc which are not trivial. Recursing
>>  "make;sudo make install" through all parts of xlibs, xapps, etc. will
>>  not account for these dependencies. As far as I can tell these
>>  dependencies are only embodied in the Imakefiles.
>
>I can express these dependencies in a top-level totally portable [1] Makefile
>without much hassle.  Would that be good enough?

Yes, that would be ideal. We have heard that the modular tree is now 
complete and self sufficient. Exercising a top level Makefile on 
clean systems without anything in /usr/X11R6 would be a good acid 
test of this claim.

At 2:15 PM +1100 1/26/05, Daniel Stone wrote:
>I know because I have bootstrapped my way with a machine with no 
>monolithic bits
>ever having been intsalled on it, up to the top of the Debrix stack, 
>cleanly.  I
>still have some Debrix bits to commit.
>
>I did this not because it's an even halfway worthwhile use case, but to prove
>that the modular tree can be used totally independently of the 
>monolithic tree,
>and is ready to make a clean break.

Make this easy to do so others can duplicate in their own environment 
and you will win converts. :-)

At 2:15 PM +1100 1/26/05, Daniel Stone wrote:
>Building your own UNIX system from scratch or whatever is a totally
>uninteresting use case, and not one I care anywhere near enough about to start
>compromising core goals for.  If you're making your OS, life is tough.  There
>are these difficult things with C libraries, compilers, and bootstrapping the
>entire thing, that suck a lot more than building X from bottom to 
>top.  There's
>a reason why everyone uses distributions.

All the same, I think everyone would agree that an OS that can not 
build itself from scratch without Herculean effort is seriously 
flawed. The same is true for X.Org. We need to provide a top level 
way to build an X11 tree to be credible in a new modular environment. 
Sure, there are lots of ways to skin a cat, but it matters when you 
are the cat. Lets at least have one solution which is supported and 
documented.

At 2:15 PM +1100 1/26/05, Daniel Stone wrote:
>(For what it's worth, I don't see the use case for building the 
>*entire* X stack
>  again either as a developer, because I can't think of anything short of
>  changing the entire wire protocol that would impact everything.)

Believe it or not, there are *nix OS'es out there which have 
excellent windowing systems that do not include X11 by default.

--Torrey