gcc-2.95.2 rough plan

gcc-2.95.2 rough plan

[Jeffrey A Law]

Just to let folks know what the rough plan is for gcc-2.95.2.

I would like to have gcc-2.95.2 hit the streets somewhere between Oct 1 and
Oct 15.  The plan (like all our minor releases) is to just fix serious bugs
that have simple and safe bugfixes.

It is very important that the proposed fixes be kept simple.  Testing before
a minor release is limited; therefore, we only want to install fixes that we
know to be very safe.

Maintainers are free to check in changes to the gcc-2.95 branch in areas of
the compiler they maintain within the guidelines noted above.

I do ask that anyone checking in changes to the gcc-2.95 branch send me 
a very short (1-2 line) description of the problem that is solved by the
change (for the release notes).

Comments and suggestions are always encouraged!


Thanks,

jeff

Re: gcc-2.95.2 rough plan

[John Vickers]

Hello.

A possible candidate for 2.95.2:

The patch:

Fix to output_strlen_unroll - avoids compiler core dump

( http://egcs.cygnus.com/ml/gcc-patches/1999-08/msg01007.html ),
which was approved by RTH,  was noted as committed to the mainline
only, although it fixed a bug apparently present at least as far back
as egcs 1.1.x.

John.

Re: gcc-2.95.2 rough plan

[Jeffrey A Law]

  In message < 37D4E9EF.926D6E7B@PaceMicro.com >you write:
  > 
  > Hello.
  > 
  > A possible candidate for 2.95.2:
  > 
  > The patch:
  > 
  > Fix to output_strlen_unroll - avoids compiler core dump
  > 
  > ( http://egcs.cygnus.com/ml/gcc-patches/1999-08/msg01007.html ),
  > which was approved by RTH,  was noted as committed to the mainline
  > only, although it fixed a bug apparently present at least as far back
  > as egcs 1.1.x.
Already installed on the branch.
jeff

Re: gcc-2.95.2 rough plan

[craig]

>I would like to have gcc-2.95.2 hit the streets somewhere between Oct 1 and
>Oct 15.  The plan (like all our minor releases) is to just fix serious bugs
>that have simple and safe bugfixes.

I've reviewed most of the g77-related bugs/patches that I've got on
file, and none of them look like they deserve to be dealt with in
2.95.anything.

        tq vm, (burley)

Re: gcc-2.95.2 rough plan

[Jeffrey A Law]

  In message < 19990907213131.28538.qmail@deer >you write:
  > >I would like to have gcc-2.95.2 hit the streets somewhere between Oct 1 an
  > d
  > >Oct 15.  The plan (like all our minor releases) is to just fix serious bug
  > s
  > >that have simple and safe bugfixes.
  > 
  > I've reviewed most of the g77-related bugs/patches that I've got on
  > file, and none of them look like they deserve to be dealt with in
  > 2.95.anything.
Have you seen any backend bugs that you're hitting regularly that someone
should take a look at?  I haven't been able to follow the bug reports as 
closely
as I would like to over the last couple months, so I'm not up to speed on any
backend problems g77 users are triggering regularly.

jeff

Re: gcc-2.95.2 rough plan

[craig]

>Have you seen any backend bugs that you're hitting regularly that someone
>should take a look at?  I haven't been able to follow the bug reports as 
>closely
>as I would like to over the last couple months, so I'm not up to speed on any
>backend problems g77 users are triggering regularly.

Hmm, I think I emailed about one or two over the past month or so,
but haven't really kept track.  I don't recall anything startling,
except for that floor() bug on some Alphas, which I assume was
fixed in time for 2.95.

        tq vm, (burley)

Re: gcc-2.95.2 rough plan

[Richard Henderson]

On Tue, Sep 07, 1999 at 11:23:17PM -0000, craig@jcb-sc.com wrote:
> except for that floor() bug on some Alphas, which I assume was
> fixed in time for 2.95.

That was a kernel bug.


r~

Re: gcc-2.95.2 rough plan

[Joel Klecker]

I'd like some comments on the following post to gcc-bugs:
< http://egcs.cygnus.com/ml/gcc-bugs/1999-08/msg00817.html >.

It's about the fact that gcc on i386 seems to default to 
-march=pentium even when configured as i386-*-*, this has caused us 
problems in Debian.
-- 
Joel Klecker (aka Espy)                    Debian GNU/Linux Developer
<URL: mailto:jk@espy.org >                 <URL: mailto:espy@debian.org >
<URL: http://web.espy.org/ >               <URL: http://www.debian.org/ >

gcc-2.95.2 rough plan

[Jeffrey A Law]

Just to let folks know what the rough plan is for gcc-2.95.2.

I would like to have gcc-2.95.2 hit the streets somewhere between Oct 1 and
Oct 15.  The plan (like all our minor releases) is to just fix serious bugs
that have simple and safe bugfixes.

It is very important that the proposed fixes be kept simple.  Testing before
a minor release is limited; therefore, we only want to install fixes that we
know to be very safe.

Maintainers are free to check in changes to the gcc-2.95 branch in areas of
the compiler they maintain within the guidelines noted above.

I do ask that anyone checking in changes to the gcc-2.95 branch send me 
a very short (1-2 line) description of the problem that is solved by the
change (for the release notes).

Comments and suggestions are always encouraged!


Thanks,

jeff

Re: gcc-2.95.2 rough plan

[Jeffrey A Law]

  In message < 19990907213131.28538.qmail@deer >you write:
  > >I would like to have gcc-2.95.2 hit the streets somewhere between Oct 1 an
  > d
  > >Oct 15.  The plan (like all our minor releases) is to just fix serious bug
  > s
  > >that have simple and safe bugfixes.
  > 
  > I've reviewed most of the g77-related bugs/patches that I've got on
  > file, and none of them look like they deserve to be dealt with in
  > 2.95.anything.
Have you seen any backend bugs that you're hitting regularly that someone
should take a look at?  I haven't been able to follow the bug reports as 
closely
as I would like to over the last couple months, so I'm not up to speed on any
backend problems g77 users are triggering regularly.

jeff

Re: gcc-2.95.2 rough plan

[Joel Klecker]

I'd like some comments on the following post to gcc-bugs:
< http://egcs.cygnus.com/ml/gcc-bugs/1999-08/msg00817.html >.

It's about the fact that gcc on i386 seems to default to 
-march=pentium even when configured as i386-*-*, this has caused us 
problems in Debian.
-- 
Joel Klecker (aka Espy)                    Debian GNU/Linux Developer
<URL: mailto:jk@espy.org >                 <URL: mailto:espy@debian.org >
<URL: http://web.espy.org/ >               <URL: http://www.debian.org/ >

Re: gcc-2.95.2 rough plan

[Richard Henderson]

On Tue, Sep 07, 1999 at 11:23:17PM -0000, craig@jcb-sc.com wrote:
> except for that floor() bug on some Alphas, which I assume was
> fixed in time for 2.95.

That was a kernel bug.


r~

Re: gcc-2.95.2 rough plan

[craig]

>I would like to have gcc-2.95.2 hit the streets somewhere between Oct 1 and
>Oct 15.  The plan (like all our minor releases) is to just fix serious bugs
>that have simple and safe bugfixes.

I've reviewed most of the g77-related bugs/patches that I've got on
file, and none of them look like they deserve to be dealt with in
2.95.anything.

        tq vm, (burley)

Re: gcc-2.95.2 rough plan

[craig]

>Have you seen any backend bugs that you're hitting regularly that someone
>should take a look at?  I haven't been able to follow the bug reports as 
>closely
>as I would like to over the last couple months, so I'm not up to speed on any
>backend problems g77 users are triggering regularly.

Hmm, I think I emailed about one or two over the past month or so,
but haven't really kept track.  I don't recall anything startling,
except for that floor() bug on some Alphas, which I assume was
fixed in time for 2.95.

        tq vm, (burley)

Re: gcc-2.95.2 rough plan

[Jeffrey A Law]

  In message < 37D4E9EF.926D6E7B@PaceMicro.com >you write:
  > 
  > Hello.
  > 
  > A possible candidate for 2.95.2:
  > 
  > The patch:
  > 
  > Fix to output_strlen_unroll - avoids compiler core dump
  > 
  > ( http://egcs.cygnus.com/ml/gcc-patches/1999-08/msg01007.html ),
  > which was approved by RTH,  was noted as committed to the mainline
  > only, although it fixed a bug apparently present at least as far back
  > as egcs 1.1.x.
Already installed on the branch.
jeff

Re: gcc-2.95.2 rough plan

[John Vickers]

Hello.

A possible candidate for 2.95.2:

The patch:

Fix to output_strlen_unroll - avoids compiler core dump

( http://egcs.cygnus.com/ml/gcc-patches/1999-08/msg01007.html ),
which was approved by RTH,  was noted as committed to the mainline
only, although it fixed a bug apparently present at least as far back
as egcs 1.1.x.

John.

RE: gcc-2.95.2 rough plan

[Billinghurst, David (RTD)]

single precision complex on 64 bit machines?

> -----Original Message-----
> From:	Jeffrey A Law [SMTP:law@cygnus.com]
> Sent:	Wednesday, 8 September 1999 8:08
> To:	craig@jcb-sc.com
> Cc:	gcc@gcc.gnu.org
> Subject:	Re: gcc-2.95.2 rough plan 
> 
>   In message < 19990907213131.28538.qmail@deer >you write:
>   > >I would like to have gcc-2.95.2 hit the streets somewhere between Oct
> 1 an
>   > d
>   > >Oct 15.  The plan (like all our minor releases) is to just fix
> serious bug
>   > s
>   > >that have simple and safe bugfixes.
>   > 
>   > I've reviewed most of the g77-related bugs/patches that I've got on
>   > file, and none of them look like they deserve to be dealt with in
>   > 2.95.anything.
> Have you seen any backend bugs that you're hitting regularly that someone
> should take a look at?  I haven't been able to follow the bug reports as 
> closely
> as I would like to over the last couple months, so I'm not up to speed on
> any
> backend problems g77 users are triggering regularly.
> 
> jeff

Re: gcc-2.95.2 rough plan

[Jeffrey A Law]

  In message <A9E96A79C068D211A6A90000C07BDF0D36947B@crtsmail.crts.techaust.rio
tinto.com.au>you write:
  > single precision complex on 64 bit machines?
Unlikely.

There are significant problems with the internal API GCC uses to manipulate
complex values.  It is unlikely we can fix the API in a safe enough manner
for gcc-2.95.x.


jeff

Re: gcc-2.95.2 rough plan

[Jeffrey A Law]

  In message <A9E96A79C068D211A6A90000C07BDF0D36947B@crtsmail.crts.techaust.rio
tinto.com.au>you write:
  > single precision complex on 64 bit machines?
Unlikely.

There are significant problems with the internal API GCC uses to manipulate
complex values.  It is unlikely we can fix the API in a safe enough manner
for gcc-2.95.x.


jeff

RE: gcc-2.95.2 rough plan

[Billinghurst, David (RTD)]

single precision complex on 64 bit machines?

> -----Original Message-----
> From:	Jeffrey A Law [SMTP:law@cygnus.com]
> Sent:	Wednesday, 8 September 1999 8:08
> To:	craig@jcb-sc.com
> Cc:	gcc@gcc.gnu.org
> Subject:	Re: gcc-2.95.2 rough plan 
> 
>   In message < 19990907213131.28538.qmail@deer >you write:
>   > >I would like to have gcc-2.95.2 hit the streets somewhere between Oct
> 1 an
>   > d
>   > >Oct 15.  The plan (like all our minor releases) is to just fix
> serious bug
>   > s
>   > >that have simple and safe bugfixes.
>   > 
>   > I've reviewed most of the g77-related bugs/patches that I've got on
>   > file, and none of them look like they deserve to be dealt with in
>   > 2.95.anything.
> Have you seen any backend bugs that you're hitting regularly that someone
> should take a look at?  I haven't been able to follow the bug reports as 
> closely
> as I would like to over the last couple months, so I'm not up to speed on
> any
> backend problems g77 users are triggering regularly.
> 
> jeff