Re: [Fwd: native hppa2.0w-hp-hpux11.11]

Re: [Fwd: native hppa2.0w-hp-hpux11.11]

[Michael Elizabeth Chastain]

> I think the most useful thing to find is a copy of the relevant ABI. 
> With that it should be possible to revamp the code, bringing it up to 
> scratch.

I'm not close to that yet.

. readline does not build with gcc.  I decided to tackle this first.
  I'm creating some patches for the standalone readline and I will
  submit them to Chet Ramey, the readline maintainer.  If he likes them,
  then I will ask Elena to put them in our local readline as well.

. I have to decide which compilers to build gdb with.  There are
  four compilers to choose from: gcc, the bundled cc, the unbundled
  ansi c compiler, and the aC++ compiler.

  The bundled cc isn't even installed on the HP test drive machines
  so I expect that I will just punt on that.  I think that all the
  others need to work though.

. I have to decide which compilers to build test programs with,
  and figure out which debug formats make sense to test.  Also there is
  32-bit versus 64-bit and pre-standard-c++-library versus
  standard-c++-library.

. I have to do something about all the little cheats in migchain and
  migbat where they are not really set up for multiple architectures
  yet.  I don't really have to fix them, but I have to get them to the
  point where I can do work with them.

And here's a big one:

. What do we do about hpux 10.20?

  There's no hpux 10.20 in the HP test drive pool.  HP has a good reason
  for this: they discontinued hpux 10.20 on 2002-06-30, and obsoleted it
  on 2003-06-30!

    http://h21007.www2.hp.com/dspp/tech/tech_TechDocumentDetailPage_IDX/1,1701,5143,00.html

  All of hpux10* is no longer supported by HP.  So we have no maintainer
  and no machine.  But we do have a recent PR, PR gdb/1411.

  I would like to give hppa*-hp-hpux10* the boot.  Put it through the
  obsoletion process, and then remove any test code which is
  hpux10-specific, and then not worry about breaking hpux10 while I am
  fixing hpux11.

  What do you think of that?

Michael C

Re: [Fwd: native hppa2.0w-hp-hpux11.11]

[Joel Brobecker]

> . readline does not build with gcc.

Reminds me something familiar, although the problem I encountered
were only when building for hppa64. See: 
http://sources.redhat.com/ml/gdb-patches/2003-01/msg00753.html.

What is the error you are getting?

> . What do we do about hpux 10.20?

In my mind, hpux-10.20 was already no longer supported by GDB :-).
We don't have a machine running that OS either, so we can't help
on this front. 

-- 
Joel

Re: [Fwd: native hppa2.0w-hp-hpux11.11]

[Andrew Cagney]

> I'm getting a similar error to you, with mbrtowc (not mbsrtowc).
> 
> Did you get a response back from bash-maintainers on your patch?
> 
> Our solutions are different.  Your solution is to add more checks
> so that hpux will properly not be multi-byte.  My solution is
> to add a README.hpux file which tells the user to do this:
> 
>   CPPFLAGS=-D_XOPEN_SOURCE=500 configure
> 
> Come to think of it, both parts are valid.  The autoconfiguration stuff
> should be set up the way you like, and then, if people want I18N support
> on HP, _XOPEN_SOURCE should enable it.
> 
> Here's the HP reference on this:
> 
>   http://www.docs.hp.com/hpux/onlinedocs/5187-3605/00/00/1-con.html
> 
>> In my mind, hpux-10.20 was already no longer supported by GDB :-).
> 
> 
> Well, that makes it 2-0.

(I'll not object) so ....

Andrew

Re: [Fwd: native hppa2.0w-hp-hpux11.11]

[Michael Elizabeth Chastain]

I'm getting a similar error to you, with mbrtowc (not mbsrtowc).

Did you get a response back from bash-maintainers on your patch?

Our solutions are different.  Your solution is to add more checks
so that hpux will properly not be multi-byte.  My solution is
to add a README.hpux file which tells the user to do this:

  CPPFLAGS=-D_XOPEN_SOURCE=500 configure

Come to think of it, both parts are valid.  The autoconfiguration stuff
should be set up the way you like, and then, if people want I18N support
on HP, _XOPEN_SOURCE should enable it.

Here's the HP reference on this:

  http://www.docs.hp.com/hpux/onlinedocs/5187-3605/00/00/1-con.html

> In my mind, hpux-10.20 was already no longer supported by GDB :-).

Well, that makes it 2-0.  But Andrew has seven votes.  :)

Michael C

[Fwd: native hppa2.0w-hp-hpux11.11]

[Andrew Cagney]

[-- Attachment #1: Type: text/plain, Size: 168 bytes --]

Michael,

I think the most useful thing to find is a copy of the relevant ABI. 
With that it should be possible to revamp the code, bringing it up to 
scratch.

Andrew

[-- Attachment #2: native hppa2.0w-hp-hpux11.11 --]
[-- Type: message/rfc822, Size: 3345 bytes --]

From: mec.gnu@mindspring.com (Michael Elizabeth Chastain)
To: gdb-testers@sources.redhat.com
Subject: native hppa2.0w-hp-hpux11.11
Date: Mon,  8 Dec 2003 17:20:35 -0500 (EST)
Message-ID: <20031208222035.CD0DD4B412@berman.michael-chastain.com>

After a bunch of script twiddling and a brief fight with readline,
I got some numbers somehow.

I haven't decided what to do next.

Michael C

=== metadata ===

  target = native
  host = hppa2.0w-hp-hpux11.11
  gcc = 3.3.2
  gdb = 6.0
  gas = 2.14
  ld = vendor
  libc = vendor
  gformat = stabs+
  glevel = 2

=== gdb Summary ===

  # ERROR                         16
  # WARNING                       5
  # of expected passes            9107
  # of unexpected failures        612
  # of unexpected successes       91
  # of expected failures          75
  # of known failures             16
  # of unresolved testcases       13
  # of untested testcases         3
  # of unsupported tests          11