This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[tree-ssa/wwwdocs] Add merge announcement
- From: Diego Novillo <dnovillo at redhat dot com>
- To: "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>
- Date: Wed, 25 Feb 2004 09:48:43 -0500
- Subject: [tree-ssa/wwwdocs] Add merge announcement
- Organization: Red Hat Canada
Adds the merge announcement and branch closure. Verified with
w3.validator.org
Diego.
Index: index.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/projects/tree-ssa/index.html,v
retrieving revision 1.28
diff -d -c -p -r1.28 index.html
*** index.html 30 Dec 2003 18:47:00 -0000 1.28
--- index.html 25 Feb 2004 14:41:40 -0000
***************
*** 9,14 ****
--- 9,15 ----
<h2>Table of Contents</h2>
<ul>
+ <li><a href="#news">Latest news (last updated: 2003-02-25)</a></li>
<li><a href="#intro">Introduction</a></li>
<li><a href="#documentation">Documentation</a></li>
<li><a href="#contributing">Contributing</a></li>
***************
*** 22,27 ****
--- 23,40 ----
</ul>
<hr />
+ <h2><a name="news">Latest News</a></h2>
+
+ <p>The branch will be merged into mainline to become part of the next
+ release of GCC. As such, the branch is now <em>closed</em> to new
+ development.</p>
+
+ <p>The stabilization process will be driven by the merge criteria
+ stated in <a
+ href="http://gcc.gnu.org/ml/gcc/2004-02/msg01294.html">http://gcc.gnu.org/ml/gcc/2004-02/msg01294.html</a>.
+ We expect to merge the branch into mainline by mid to late April.</p>
+
+ <hr />
<h3><a name="intro">Introduction</a></h3>
<p>The goal of this project is to build an optimization framework for trees
*************** href="http://people.redhat.com/dnovillo/
*** 57,62 ****
--- 70,78 ----
documentation</a> for the Tree SSA infrastructure is generated from the
source code using Doxygen. The documentation is updated daily.</p>
+ <p>Internal design documentation is also available in the texinfo
+ documentation (<code>doc/gccint.info</code> in GCC's build directory).</p>
+
<p>A high-level overview of GENERIC/GIMPLE and the SSA implementation may
be found in the <a
href="http://www.linux.org.uk/~ajh/gcc/gccsummit-2003-proceedings.pdf">
*************** Diego Novillo, Sebastian Pop, Graham Sto
*** 93,139 ****
<hr />
<h3><a name="stability">Branch stability</a></h3>
! <p>We are currently working on stabilizing the branch to make it
! possible to merge into mainline for the 3.5 release. The goal is
! very simple: <em>make the branch better than mainline</em>. For
! details, see the <a href="#shortterm">short term goals</a>
! section.</p>
!
! <p>Every patch submitted for review must comply with the
! following requirements:</p>
!
! <ol>
! <li>The patch must be complete, not a work in progress.</li>
! <li>If the patch is too large or depends on other fixes, submit
! them separately.</li>
! <li>Patches must have been bootstrapped with <em>all</em> the
! default front ends and regression tested on at least one of
! the main native architectures against a pristine tree, with
! <em>no</em> new regressions on any testsuite.</li>
! <li>If the tree changes just before you are about to commit your
! patch, you <em>must</em> go through the bootstrap and
! regression test cycle once again.</li>
! </ol>
<p>Patches that break default bootstraps or introduce new
regressions will be removed (if a fix is not immediately
obvious).</p>
- <p>Also, for new analyses and transformations, please include a
- reference to the paper and/or book where you are getting the
- algorithm from. If it's your own research work, include a
- Technical Report, Thesis or Paper reference for it.</p>
-
<p>There are regular mainline merges about 2 or 3 times a month. The
latest merge tag is always added to GCC's version string. A merge may
be postponed if there is major breakage in mainline.</p>
- <p>Work from the branch will be incorporated into mainline only
- when we can prove improvements in code quality, compile times and
- memory utilization. We should be able to either replace or
- simplify some of the existing transformations currently done on
- RTL.</p>
-
<hr />
<h2><a name="gimple">GENERIC and GIMPLE</a></h2>
--- 109,127 ----
<hr />
<h3><a name="stability">Branch stability</a></h3>
! <p>Every patch submitted for review must either fix a PR or adress one
! of the issues mentioned in the <a
! href="http://gcc.gnu.org/ml/gcc/2004-02/msg01294.html">merge criteria
! document</a>.</p>
<p>Patches that break default bootstraps or introduce new
regressions will be removed (if a fix is not immediately
obvious).</p>
<p>There are regular mainline merges about 2 or 3 times a month. The
latest merge tag is always added to GCC's version string. A merge may
be postponed if there is major breakage in mainline.</p>
<hr />
<h2><a name="gimple">GENERIC and GIMPLE</a></h2>
*************** existing DejaGNU testing framework.</p>
*** 303,336 ****
possible improvement, and planned analyses and optimizations.
Suggestions for other passes and volunteers to help finish the
different passes are welcome.</p>
-
- <h3><a name="shortterm">Short term goals</a></h3>
-
- <dl>
- <dt><em>Speed up the new infrastructure</em></dt>
- <dd>There are still many things that we can do faster. The compile time
- performance of GCC should improve, not degrade. A lot of work is
- going on in this area.</dd>
-
- <dt><em>Fix Bugzilla reports</em></dt>
- <dd>A new Bugzilla category has been created for <a
- href="http://gcc.gnu.org/bugzilla/">bug reports</a> against the branch
- (search for version <code>tree-ssa</code>). They should all be fixed or at
- least analyzed to determine whether they are merge blockers.</dd>
-
- <dt><em>Improve compile time memory utilization</em></dt>
- <dd>We have several memory footprint and locality problems. Some
- are caused by new data structures that have still not been
- optimized for size, others are caused by the tendency of creating
- many more trees than the mainline compiler.</dd>
-
- <dt><em>Improve run time performance</em></dt>
- <dd>The quality of generated code has improved, but it still lags
- behind mainline. We use SPEC CPU2000 and SPEC CPU95 to track
- performance. Performance results comparing the branch against
- mainline using freely available suites would be most
- welcome.</dd>
- </dl>
<h3>Medium and long term goals</h3>
--- 291,296 ----