This is the mail archive of the
mailing list for the GCC project.
Criteria for performing Ada-included bootstraps
- From: "Zack Weinberg" <zack at codesourcery dot com>
- To: gcc at gcc dot gnu dot org
- Date: Fri, 13 Feb 2004 11:31:06 -0800
- Subject: Criteria for performing Ada-included bootstraps
- References: <firstname.lastname@example.org><Pine.BSF.email@example.com>
Gerald Pfeifer <firstname.lastname@example.org> writes:
> On Sun, 25 Jan 2004, Zack Weinberg wrote:
>> However, it is my understanding that breaks that affect only Ada are
>> *not* considered critical, and in fact that there is no obligation for
>> anyone to attempt to build the Ada front end.
> In principle there *is* such an obligiation, though we have not been
> enforcing it in the past for several reasons.
> However, now that the Ada frontend is maintained very actively and
> some long-standing installation/"portability"/... issues are getting
> resolved, I expect this to change in the long term.
In the long term that is fine, however I think it is premature at this
point and will still be premature for some time to come.
I cannot speak for anyone other than myself. But, Laurent asked me
under what circumstances I would *personally* be willing to turn Ada
back on when testing my own patches. I didn't want to sidetrack that
thread, but it is a fair question and I will attempt to answer it now.
First, work on the separate libada must be complete, including support
for multilibs. This is well under way and I have confidence it will
be done soon.
Second, the Ada maintainers have said in the past that they
deliberately break source compatibility between releases (such that
GCC 3.x with Ada is only guaranteed to be buildable with GCC 3.(x-1)).
This is not acceptable. It is not practical for me to have more than
one Ada bootstrap compiler installed per test platform. There must be
guaranteed bidirectional source compatibility between all 3.x branches
of GCC, such that it is possible to build any 3.x version of GNAT with
any other 3.x version of GNAT. Ideally, one-way source compatibility
as far back as GNAT 2.8.1 would be maintained indefinitely; I have no
need to build that version, but it may be the only bootstrap compiler
I can get my hands on for some platforms.
Since the Ada maintainers have said in the past that they deliberately
break source compatibility, I will not accept anything less than a
publicly stated change of policy, plus patches to fix any
incompatibilities that currently exist (or test results that
demonstrate that there are none) as evidence that the branches are in
Third, I want to see Ada development carried out according to the same
rules as all other GCC development. That means: all patches are
submitted individually, by their original authors, and reviewed
publicly; all bug fixes come with test cases; all new features come
with documentation; and so on. Until and unless this happens I do not
consider the Ada team to be cooperating with the rest of us and
therefore I see no reason to reciprocate.
These are my personal opinions, which may or may not reflect the views
of any other person or entity.
These are absolutely non-negotiable. I will ignore all attempts to
persuade me to change my mind, and I will oppose any proposal to
mandate Ada-included bootstraps until these conditions are met.
To the extent of my time and ability I am willing to assist in the
removal of technical hurdles to the implementation of these
conditions, provided that there is agreement in principle.