Re: build/1411: Build of gdb-6.0 on hppa1.1-hp-hpux10.20 fails

[mec dot gnu at mindspring dot com (Michael Elizabeth Chastain)]

The following reply was made to PR build/1411; it has been noted by GNATS.

From: mec.gnu@mindspring.com (Michael Elizabeth Chastain)
To: dave@hiauly1.hia.nrc.ca
Cc: gdb-gnats@sources.redhat.com
Subject: Re: build/1411: Build of gdb-6.0 on hppa1.1-hp-hpux10.20 fails
Date: Sun, 14 Dec 2003 03:06:54 -0500 (EST)

 Dave,
 
 Calm down.
 
 > Are you now going to consider dropping 11.00 and 11.11 support?
 
 No.  If you read the HP document that I referred to, you will see that
 HP is continuing to support 11.00 and 11.11.
 
 If you read gdb-testers, you will see that I recently reported test
 results from hppa2.0w-hp-hpux11.11.
 
 And if you read gdb-patches, you will see that I wrote, tested, and
 submitted a patch to fix a build failure on hppa2.0w-hp-hpux11.11
 in _rl_vi_change_mbchar_case.
 
 I would be pleased if you were to try that patch and report whether it
 fixes the problem on your system too.
 
 > Is HP making decisions for the gdb project?
 
 HP's decisions affect my opinions, and my opinions are part of
 the input in gdb project decisions.
 
 I can't QA platforms that are not available to me.  HP doesn't have any
 hpux 10.20 machines in their test drive program.
 
 Would you like to run the gdb test suite regularly on hpux 10.20 and
 report the results to gdb-testers?
 
 Would you like to give an account on an hpux 10.20 so that I could do
 that?
 
 > Are you applying the same standards to all OS's that are not
 > supported?
 
 Yes.  If you read the gcc mailing list, you will see that my standard
 is: if no results have been reported for three minor revisions, then I
 want to start the obsoletion process on that platform.  That's my
 personal standard, not a gdb project standard.
 
 > Do you plan to drop 11i1.5 on ia64.  It is very similar at least from an
 > OS standpoint to 11i on the PA?  It was obsoleted before 10.20.
 
 Yes, I do.  I don't have access to an 11i v1.5 system and I'm not likely
 to get one through HP.  I do have access to 11i and 11i v1.6.
 
 > It appears from your response that PA-RISC support within the
 > gdb project is seriously threatened.
 
 You are entirely wrong.  Please read my recent messages in gdb@,
 gdb-patches@, and gdb-testers@, and then judge whether I am working to
 improve PA-RISC support or destroy it.
 
 I want to remove support hppa*-hp-hpux10.*, yes.  And I want to
 support hppa*-hp-hpux11.* at the same level I support i686-pc-linux-gnu.
 
 > A bug report isn't the place to discuss target support issues.
 
 Yes, it is.  You reported a bug on a target that I want to deprecate.
 It's germane to mention that.
 
 > 6.0 was released with nominal support for hppa1.1-hp-hpux10.20.
 
 Well, gdb hppa1.1-hp-hpux10.20 can continue to have nominal support
 where nobody actually works on it.
 
 Or it can have whatever support that you or anybody else wants to
 provide.
 
 And gdb hppa2.0w-hp-hpux11.11 can continue to have nominal support
 where nobody actually works on it.
 
 Or it can have my support, where I actually build gdb and run the test
 suite several times per month; report and analyze the results; file PR's
 for every regression since the last release; and even, occassionally,
 fix some of the PR's.
 
 Or it can have whatever support that you or anybody else wants to
 provide.
 
 The gotcha is that if I work on hpux 11, I am going to push hard for
 obsoletion of hpux 10.
 
 I am sick of gdb having a long list of targets that used to work in some
 indefinite past, but that are no longer even BUILT, let alone tested,
 let alone the tests analyzed.  I am attacking this problem in two ways.
 First, I am picking up platforms one at a time and bringing up the
 support level on them.  Second, I am looking to broom out things that we
 can't do.
 
 I can't fix any problems with gdb on hpux 10.20.  I can't even promise
 not to make it worse.
 
 > GCC first deprecates a target, then removes it in the following release
 > if sufficient interest hasn't been expressed to revive it.  This seems
 > like a good model.  GCC still supports 10.X, so you will be leading GCC
 > if you drop support for 10.X at this point in time.
 
 I spoke unclearly.  My plan is: when Andrew comes back from his trip,
 I will propose to obsolete the hppa*-hp-hpux10.* targets, using the
 normal gdb obsoletion procedure.  If anyone would like to volunteer to
 support this platform then that would be great.
 
 And we may already be leading gcc, because gdb already doesn't build
 on native hppa1.1-hp-hpux10.20.  I dunno if gcc 3.3.2 builds on that
 platform or not.
 
 Michael C