Re: [Chicken-users] CMake problem on Linux should be solved

[felix winkelmann]

On 9/8/06, Brandon J. Van Every <address@hidden> wrote:
> >
> > Don't make a political thing out of this, Brandon. We are talking
> > about the development repository of an actively maintained open source
> > project. People involved do make mistakes, bugs appear and get fixed
> > (hopefully). If I made a change to Makefile.am without fixing the
> > appropriate
> > section in CMakeLists.txt right away then it was simply by mistake.
>
> This is not a confidence I would have previously had, but I will take
> your word for it in the future.

Good.

> I'm skating eviction for this fucking project.  Do you have any idea how
> much real income I've lost on this?

Why you did your contribution and how that works out with your financial
situation is not of importance here. You've done good work, work that
will make Chicken a better project, and that is all that counts.

But I won't let anybody get away with declaring that I'm not trying
to do the utmost possible to incorporate valuable contributions,
or that I even break things deliberately, even by neglection.

> This is a difficult stage, because I can't actually make you stop adding
> new features.  You said privately, somewhere like 1..2 weeks ago, that
> you had broken too many things, wanted a stabilization period for
> Chicken, and that you now saw CMake as a top priority.  That resurrected
> my flagging morale.  Previously, I was starting to think that no matter
> how much I fix Chicken, there will be double the number of things
> waiting around the corner to undo it again.  So tonight, when I pulled 4
> significant looking patches, with substantial looking Autoconf changes
> for cross-compiling among them, and things broke in several places, I
> was not happy.  It did not look like you had taken me seriously, about
> the impact it has on my time and finances.  But you say it's just a
> Darcs problem, that you handled what was supposed to be handled, and it
> just didn't make it in.  So this leads me to a different conclusion...

Yeah, I'm quick with changes and sometimes commit experimental code,
but this is how Chicken is developed, and the pace has made it one of
the fastest growing and improving Scheme/Lisp implementations
in existance.

> ...we are not communicating with each other about what we're doing.
> Darcs is a great enabler for that.  It can in principle let our patches
> fly right past each other.  It has worked previously; it is not working
> anymore.  We need to start talking about what we're doing, because the
> basis of trust is somewhat eroded.

But not due to technical issues. Writing and build-system is just a hard
thing to do and very complicated, especially with a compiler that
bootstraps itself. This silly, tiny issue about parsing the version string
in CMakeLists.txt was just a coincidence.

Yes, I want more stability, but I also want a system that works well - there
is naturally a trade off and fast development has priority (for me, at least).

In fact, I think we've done quite well.

> I am unsure what milestones you are trying to hit.  2 weeks ago, I
> thought I knew.  But we never made any kind of formal agreement on date,
> time, content, or anything like that.  So perhaps you think you're
> hitting a milestone 1 month from now.  Or 2 months from now.  I really
> don't know.  I worry that we don't actually share goals.  That in
> practice, you'll just keep designing for Autoconf to get
> cross-compilers, and CMake will become the biggest "eh" I've ever
> written.  (But at least a BSD licensed "eh.")

There is no real cross-compilation support. The stuff modified in the
autoconf files was just the start of an experiment - far from being finished.
Adding that bit to CMakeLists.txt would be trivial if I just new more about
CMake (or if the documentation would be better...). I planned to add it
anyway.

> We need to agree what the milestones are.  Or, you need to tell me what
> the milestones are gonna be, if you're in a more forceful mood about
> it.  Then I can decide what I'm willing to do about them.

I have no milestones, besides getting 2.5 out as soon as I find some time
doing testing which includes booting up that despised Windows notebook...
There are no new features planned, and whatever I muddle around in
configure.in shouldn't concern you, since it is simply not ready.

> >
> > This is the last warning.
> >
>
> I'm not at the stage of delivering ultimatums.  Be advised that I'm
> capable of them, however.  In the worst case, I do have the willingness
> to write off my past 2 years as a failed experiment in open source
> business models.  There has to be a point at which I can just use the
> results of my work, and not just end up with more work.  I know I won't
> be repeating the kind of open source investment I've made into Chicken.
> For instance, regarding Chickenizing G3D, I am very cautious.  Not
> interested in another 10 months.

You are free to do whatever you want, of course. The CMake support is
very good, and I appreciate that and will go out of my way to acknowledge
that properly when the next version will be released. If you could just get that
idea out of your head that I'm trying to break your work, or whatever. Also, you
should perhaps keep your private interests in supporting Chicken out of the
general discussion, as that is something you have to work out for yourself.
If you want to contribute, contribute. Don't try to get your personal
motiviations
into the discussion, please. I've put 6 years of my free hacking time into
this (nothing else), and nobody should give a damn about my personal
motivations or how often I get fed up with it, or how it affects my day job.

Help us with your experience, if you want. But don't complain, as we all are
doing our best.


cheers,
felix

--
http://galinha.ucpel.tche.br:8081/blog/blog.ssp