Re: GCC 3.0 Release Criteria

[Mark Mitchell <mark at codesourcery dot com>]

>>>>> "Martin" == Martin v Loewis <martin@loewis.home.cs.tu-berlin.de> writes:

    >> I would also like to see one of the windows ports, preferably
    >> Cygwin port, as a second-tier platform. I'll do the necessary
    >> testing (I still need to spend a few hours and get the patches
    >> out of my local tree, sigh).

    Martin> I cannot understand the notion of second-tier platforms,
    Martin> at all.

Your point is valid.  However, I think there are several purposes to
having a list of second-tier platforms:

  - The community can see what is actually being worked on.  If
    we're forgetting an important platform, someone can complain.

  - The community can get a sense of what platforms GCC 3.0 is
    most likely to work on.  If you see that your platform is 
    a second-tier platform, that will at least imply that GCC
    3.0 has bootstrapped on your platform, and is likely to work
    at least somewhat.

  - It gives us a way to associate volunteers with tasks.  We're
    not at that stage yet, but it will happen soon.  When someone
    commits to sponsoring PowerPC Linux, say, then we'll have
    a place to hang that tag on the web site.

    Martin> Serious problems will delay the release if there is a
    Martin> chance that they get fixed before the release. That does
    Martin> not primarily depend on the platform being listed as
    Martin> secondary, but whether there are people that actually do
    Martin> offer to fix it, within some period of time.

Yes -- except that if the platform is primary, we'll not ship.  A
volunteer will fix the bug, or we'll just sit there.  At least for
quite some time.

I see it as vital that we not ship on, say, x86 GNU/Linux until we
know things are working very smoothly.  We'll receive a severe
backlash, from lots of people, if there are serious problems.  And, I
think there would be negative repurcussions for the GNU Project as a
whole.

For the secondary platforms, if nobody volunteers after a while, we'll
give up and ship anyhow.

If there's the odd bug or two on SunOS 4.1.4, that's too bad -- but
not nearly as damaging to the GNU Project as whole, in my opinion.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com