Looking at the overlap of the FC4 schedule with the GCC 4.0 schedule, it appears that shipping GCC 4.0 in FC4 has become viable.
In order to avoid a large duplication of bandwidth by rebuilding with GCC 4.0 between test releases, Fedora Core 4 Test 1 has now slipped two weeks to allow for integration of a GCC 4.0 snapshot, and rebuilds against it.
Fedora Core 4 Test 1 is now scheduled for March 14; the schedule at http://fedora.redhat.com/participate/schedule/ will be updated shortly.
Quick FAQs:
- Is GCC 4.0 released yet?
No. It's likely to be released around mid-April.
- Does that mean Fedora Core 4 will ship with a pre-release compiler?
We're not *that* crazy. If GCC 4.0 is delayed, we will either revert, or slip.
- What's so cool about GCC 4.0?
GCC 4.0 includes:
- new intermediate languages that allow for new/better optimizations http://gcc.gnu.org/projects/tree-ssa/
As part of this, this allows for better compile and run-time memory checking and overflow protection.
- better and more complete Java support - Fortran 95 support
- Is GCC 4.0 ABI compatbile with GCC-3.4.x?
- For C/C++, yes.
Bill
On Wed, Feb 23, 2005 at 05:06:34PM -0500, Bill Nottingham wrote:
Looking at the overlap of the FC4 schedule with the GCC 4.0 schedule, it appears that shipping GCC 4.0 in FC4 has become viable.
In order to avoid a large duplication of bandwidth by rebuilding with GCC 4.0 between test releases, Fedora Core 4 Test 1 has now slipped two weeks to allow for integration of a GCC 4.0 snapshot, and rebuilds against it.
This may be an obvious question, but how do non-x86 architectures play into this? If there issues with these architectures and GCC 4.0, will Fedora Core 4 revert to GCC 3?
I ask because I saw rumblings of some ppc32/gcc4 problems today. It may be a false alarm, but the question still stands. Feel free to add it to the FAQ :).
josh
On Wed, Feb 23, 2005 at 04:17:10PM -0600, Josh Boyer wrote:
This may be an obvious question, but how do non-x86 architectures play into this? If there issues with these architectures and GCC 4.0, will Fedora Core 4 revert to GCC 3?
I ask because I saw rumblings of some ppc32/gcc4 problems today. It may be a false alarm, but the question still stands. Feel free to add it to the FAQ :).
Would appreciate to know about any such rumblings ;). If it is reported upstream, then upstream PRs numbers, otherwise file bugs in our bugzilla. For the rebuild gcc4 needs to work on all 7 arches rawhide is built on.
Jakub
On Wed, Feb 23, 2005 at 05:24:36PM -0500, Jakub Jelinek wrote:
On Wed, Feb 23, 2005 at 04:17:10PM -0600, Josh Boyer wrote:
This may be an obvious question, but how do non-x86 architectures play into this? If there issues with these architectures and GCC 4.0, will Fedora Core 4 revert to GCC 3?
I ask because I saw rumblings of some ppc32/gcc4 problems today. It may be a false alarm, but the question still stands. Feel free to add it to the FAQ :).
Would appreciate to know about any such rumblings ;). If it is reported upstream, then upstream PRs numbers, otherwise file bugs in our bugzilla. For the rebuild gcc4 needs to work on all 7 arches rawhide is built on.
http://lkml.org/lkml/2005/2/23/119
Dave
On Wed, Feb 23, 2005 at 05:28:20PM -0500, Dave Jones wrote:
On Wed, Feb 23, 2005 at 05:24:36PM -0500, Jakub Jelinek wrote:
Would appreciate to know about any such rumblings ;). If it is reported upstream, then upstream PRs numbers, otherwise file bugs in our bugzilla. For the rebuild gcc4 needs to work on all 7 arches rawhide is built on.
Yep, that's what I was talking about. I called it a rumbling because nobody has responded yet.
josh
Just in case it was missed:
It looks like Jakub found the problem and has a fix. No worries!
On Wednesday 23 February 2005 14:53, Josh Boyer wrote:
On Wed, Feb 23, 2005 at 05:28:20PM -0500, Dave Jones wrote:
On Wed, Feb 23, 2005 at 05:24:36PM -0500, Jakub Jelinek wrote:
Would appreciate to know about any such rumblings ;). If it is reported upstream, then upstream PRs numbers, otherwise file bugs in our bugzilla. For the rebuild gcc4 needs to work on all 7 arches rawhide is built on.
Yep, that's what I was talking about. I called it a rumbling because nobody has responded yet.
josh
Hi,
Looking at the overlap of the FC4 schedule with the GCC 4.0 schedule, it appears that shipping GCC 4.0 in FC4 has become viable.
Can we take it then that OOo2 should also be included given this slipage?
TTFN
Paul
On Wed, 23 Feb 2005, Paul wrote:
Looking at the overlap of the FC4 schedule with the GCC 4.0 schedule, it appears that shipping GCC 4.0 in FC4 has become viable.
Can we take it then that OOo2 should also be included given this slipage?
Given a May 26th Absolute Freeze, it gets a lot more likely, doesn't it :) But as with all things, it depends on whether OOo upstream keeps its release schedule.
Dan
Bill Nottingham wrote:
[SNIP]
What's so cool about GCC 4.0?
GCC 4.0 includes:
new intermediate languages that allow for new/better optimizations http://gcc.gnu.org/projects/tree-ssa/
As part of this, this allows for better compile and run-time memory checking and overflow protection.
better and more complete Java support
Fortran 95 support
- The default C++ allocator is non-generic and optimized for multithreaded applications. If I recall correctly, OpenOffice.org relies on libstdc++.
http://gcc.gnu.org/ml/libstdc++/2005-02/msg00061.html
- GCC 4.0 offers the possibility of defining symbols as "hidden", minimizing the overhead of shared object dynamic loading. There's some information on how this is useful here:
http://people.redhat.com/drepper/dsohowto.pdf
A full list of the great new features is on:
http://gcc.gnu.org/gcc-4.0/changes.html
-- Pedro Lamarão
Hi all Apart from this list what all the places from where i can get the update information about FC4 if any known to you all then please give the pointer towards that.
crisppy f
On Thu, Feb 24, 2005 at 12:28:55AM -0300, Pedro Lamar?o wrote:
- GCC 4.0 offers the possibility of defining symbols as "hidden",
minimizing the overhead of shared object dynamic loading.
Well, visibility attribute have been there already in GCC 3.2.3-RH. -fvisibility=* and visibility attribute for classes that is new in upstream 4.0 has been backported to GCC 3.4.2-RH and later, so is already available in FC3/RHEL4 GCC.
Jakub
devel@lists.stg.fedoraproject.org