This is the mail archive of the crossgcc@sourceware.org mailing list for the crossgcc project.

See the CrossGCC FAQ for lots more information.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [crosstool-NG] Design discussion


On Friday 10 April 2009 04:09:42 Ladislav Michl wrote:
> On Wed, Apr 08, 2009 at 08:13:31PM -0500, Rob Landley wrote:
> > On Wednesday 08 April 2009 05:22:09 Ladislav Michl wrote:
> > > You do not need install PTXdist anywhere to start using it.
> >
> > Uh-huh.  Starting from a fresh tarball...
>
> I didn't do that in last few years, so perhaps there is something wrong
> with tarballs...

I do that every release, and I try to make quarterly releases because this 
guy's talk convinced me it was a good thing:

http://video.google.com/videoplay?docid=-5503858974016723264

> >   $ make menuconfig
> >   make: *** No rule to make target `menuconfig'.  Stop.
> >
> >   $ bin/ptxdist menuconfig
> >   ptxdist: error: PTXdist in /home/landley/ptxdist-1.0.2 is not built.
>
> ~/src/ptxdist-trunk$ grep AC_INIT configure.ac
> AC_INIT([ptxdist],[1.99.svn],[ptxdist@pengutronix.de])
>
> ...ah here it is, seems version differs and indeed there are some newer
> tarballs: http://www.pengutronix.de/software/ptxdist/download/v1.99/
> Note that 1.0.2 is maintenance release of 2 years old stuff...

Yeah, except the website called 1.0.2 the stable release and 1.99 the 
development branch.  (I tend to think of "development branch" as "today's 
source control snapshot".  I might play with it _after_ I get stable 
working.)

> > It seems more accurate to say there might be a non-obvious workaround for
> > the need to install it before using it.
>
> ~/src/ptxdist-trunk$ ./configure && make
> ~/src/ptxdist-trunk$ cd ../ptx-test/
> ~/src/ptx-test$ ../ptxdist-trunk/bin/ptxdist menuconfig
>
> Now it will bitch at you again about selected platform config etc...
> So those two years do not represent any progress for the point you are
> trying to make. Shame ;-)

Regression testing your newbie intro paths is something you have to do on a 
regular basis.  (They bit-rot, because the _developers_ never need them...)

> > > Install part is
> > > optional just in case you want to distribute it as binary tarball to
> > > your colleagues
> >
> > Isn't what we downloaded already a tarball?  The only "binary" part seems
> > to be the kconfig binaries...
> >
> > > or make (for example) debian package. However I have to admit
> > > that it is non obvious. PTXdist's kconfig is hacked a bit to handle
> > > dependencies, so if you want to express openssh dependency on openssl
> > > you do so in Kconfig file only.
> >
> > Doesn't kconfig normally track and enforce dependencies?  I thought that
> > was one of its main functions...
>
> ...and still is, I just didn't express it clear. If feature FOO depends on
> BAR, you compile both and link them together. No matter which one gets
> compiled first. Now when program FOO depends on library BAR you need to
> compile that library first to let FOO link against it which is something
> 'plain' kconfing does not handle.

So you're generating makefile dependencies from the kconfig dependencies...?

> > I'm trying to ask the "Do we really _need_ fins?  What about gas mileage?
> >  Is lead really _necessary_ in gasoline?" type questions.  Even if
> > everything (including the stuff I've written) currently gets this sort of
> > thing wrong, it should still be possible to do _better_...
>
> I'm wondering, why you (a computer guys) are always trying to compare
> software development to automotive industry without having any deeper clue
> about it.

I wouldn't call myself an automotive _expert_, but why do you assume I haven't 
got _any_ deeper clue about the industry?

Would you like me to explain how an internal combustion engine works?  Put you 
in touch with my friends in Michigan who are unemployed due to the layoffs at 
Ford and General Motors?  Give the whole spiel about how electric golf carts 
had the potential to be a standard disruptive technology naturally 
supplementing and then replacing gasoline powered vehicles but were precluded 
from going outside apartment complexes and such due to regulatory barriers 
restricting what could become "street legal" (and yes I'm aware of the power 
to weight ratio advantage inherent in fuels that use atmospheric oxygen as 
half their reaction mass, and yet laptops are the main commercial driver of 
battery technology today and they care at least as much about weight to power 
ratio).  How biodiesel sounds like a great idea until you realize that if all 
our arable land switched over to biodiesel prodution and we _stopped_eating_ 
we still wouldn't have replaced all the diesel we currently use?  Although at 
least it's a net producer of energy, as opposed to corn ethanol which is just 
a lossy storage mechanism nobody would take seriously if it wasn't subsidized 
by _stupid_ federal programs.

Should I go into the parallels between the 1970's and 2000's for the american 
auto industry, and why SUVs were the second coming of the fin-covered gas 
guzzlers of the earlier generation and the _continuing_ refusal to focus on 
fuel efficient vehicles (yes lower margin but the demand isn't nearly as 
elastic, and you _knew_ peak production of oil was coming from the drilling 
records back in the 80's, yes the oil companies lied about their reserves and 
quietly restated them in the wake of Enron/Worldcom/tyco/global crossing etc 
but that was 2001, you had seven years warning to retool.  If you think the 
EV1 was mishandled, why the heck did saturn start making SUVs when the POINT 
of saturn was to compete with Japanese and German sedans and subcompacts?  
Japan's lost decade should have been a _gift_, not an excuse to do an upward 
retreat _away_ from things like the Mini Cooper and the relaunched Volkswagen 
Feature (it's not a bug, I don't care what they say).  What did they do in 
the past 10 years that they're _proud_ of, an HV3 you can't actually take 
off-road without doing 10s of thousands of dollars of damage to the thing?  
Go ahead, put chrome and fins on it already!)

Anyway, all that's _deeply_ off topic for this list.  Email me if you're bored 
enough to want to go into it.

> Pretty please leave away those analogies as they are not 
> completly analogic and do not serve any real purpose.

Automobiles are the most complicated piece of reasonably mature technology 
that ordinary individuals use on a daily basis.  You need training and 
certification to operate them, they have extensive maintenance requirements 
their users need to be aware of (gas, oil, brake pads), but most people 
aren't mechanics and aren't expected to be, even though system failures can 
maim or kill.  Despite that, we take them for granted, expect everybody to 
learn to use them as a teenager, and lots of families have two.

If you want to know how people will think about computers once they've been 
around for 100 years, the automobile is an obvious model because very little 
else has been _around_ for 100 years.  (The light bulb and the rotary dial 
telephone started about the same era, but neither were in the same complexity 
category.  We're not expected to _operate_ either in a nontrivial way.)

Rob
-- 
GPLv3 is to GPLv2 what Attack of the Clones is to The Empire Strikes Back.

--
For unsubscribe information see http://sourceware.org/lists.html#faq


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]