Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 02 Jul 2002 17:00:17 -0700
From:      Terry Lambert <tlambert2@mindspring.com>
To:        Jan Lentfer <Jan.Lentfer@web.de>
Cc:        Jan Lentfer <jan@localhost.homeip.net>, Andrew Gallatin <gallatin@cs.duke.edu>, freebsd-alpha <freebsd-alpha@freebsd.org>
Subject:   Re: List of ports that can be compiled with compaq-cc
Message-ID:  <3D223E91.A71CC743@mindspring.com>
References:  <3D21F1C8.2010708@web.de> <15650.6127.427432.57976@grasshopper.cs.duke.edu> <3D22188C.2000603@web.de> <3D222E25.62D5E4D0@mindspring.com> <3D222EF3.8070700@web.de>

next in thread | previous in thread | raw e-mail | index | archive | help
Jan Lentfer wrote:
> >>But seriously.... How should we manage this? I'd say a website (with
> >>regestration for submitters???) where we could hold the status of all ports
> >
> >Modify the ports.mk so that it will check a flag, and, if it is
> >present and the compiler is present, have it "prefer" the Compaq
> >compiler.  Then for those ports where it works, just set the flag
> >in their Makefile.
> >
> >Allow this behaviour to be globally overridden via make.conf.
> 
> But before we could do this we would need a list of "known-to-work" ports.


No, you wouldn't.  That's why I designed it that way: so you
could incrementally add to the list of "known-to-work" ports
over time, incrementally, starting at 0 or 1 ports.

It also allows for "Yes, but..."; for example, if some port
did work with the Compaq compiler, but the code result was
somehow inferior to the non-Compaq compiler, you could add
the flags that it worked, then comment them out, with a text
comment saying why.

It also allows for the Compaq compiler to become a port
dependency for the ports best compiled with the Compaq compiler,
at a later date.

You appear to want a lot of people to install the Compaq tools
on a minority platform, and spend a lot of effort verifying
ports on your behalf, just so they can tag this fact in a
comment in the "README" or Makefile for a port, and obtain no
benefit from the effort, other than the tiny number of people
who read these files before executing "make" on them, in order
to build a given port.

With respect, the single most important thing that makes effort
happen with regard to Open Source is working code.  I've proven
this over and over again.  FreeBSD is around because 386BSD code
*worked*, thanks to my efforts with the patchkit.  OpenLDAP is
around because of the patches that I both collected into a single
whole, and made on my own, as an experiment to prove this thesis;
on its own, the University of Michigan LDAP code needed to *work*
before it gained wide acceptance.  Eric Raymond can take his
"The Cathedral and the Bazaar" voo-doo and stuff them.  I have
mathematical models.  I can *prove* my thesis over 9 Open Source
Software projects in which I have participated directly, and over
another half a dozen which I've analyzed.

If you want something that's going to result in ports available
for you, which can be compiled out-of-the-box with the Compaq
compiler, the only approach that's going to work consistantly
is going to be the approach I've described.  The approach that
you described will not result in what you say you want to have
happen.  The emergent properties of your approach is a significant
division of effort.

If you don't care about what happens, and you just want a project,
then, by all means, declare one at Sourceforge, and have it get
nowhere, just like all other non-working projects proceed to
failure after declaration without working code.

-- Terry

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3D223E91.A71CC743>