- Projects

[LrdeTools] 417: Update brasilia's settings for ICC 9.1.
by Roland Levillain 14 Feb '07

14 Feb '07

https://svn.lrde.epita.fr/svn/lrde-tools/trunk/build-farm Index: ChangeLog from Roland Levillain <roland(a)lrde.epita.fr> Update brasilia's settings for ICC 9.1. * buildfarm_worker/brasilia.fns (host_type): Set to linux-x86. Add ICC 9.1's directory to the PATH. Do not use ccache with ICC, as ICC tends to mix up with generated dependencies. (init_icc_vars): Adjust for ICC 9.1. (test_olena_tree): Add support of additional options. Disable ccache when using ICC. Use this function to test Olena 0.10a trees. Add a test tree for Olena 0.10a using the Intel C++ Compiler (icpc) version 9.1. * web/build.pl: Remove olena-0.10a and olena-1.0-icpc. Add olena-0.10a-g++-4.1, olena-0.10a-icpc-9.1 and olena-1.0-icpc-9.1. buildfarm_worker/brasilia.fns | 111 +++++++++++++----------------------------- web/build.pl | 3 + 2 files changed, 39 insertions(+), 75 deletions(-) Index: web/build.pl --- web/build.pl (revision 416) +++ web/build.pl (working copy) @@ -83,10 +83,13 @@ 'vaucanson-icc', '---', 'olena-0.10a', + 'olena-0.10a-g++-4.1', + 'olena-0.10a-icpc-9.1', 'olena-1.0-g++-3.4', 'olena-1.0-g++-4.0', 'olena-1.0-g++-4.1', 'olena-1.0-icpc', + 'olena-1.0-icpc-9.1', '---', 'ranch', '---', Index: buildfarm_worker/brasilia.fns --- buildfarm_worker/brasilia.fns (revision 416) +++ buildfarm_worker/brasilia.fns (working copy) @@ -2,6 +2,11 @@ # Used to compile Olena and Vaucanson. +host_type=linux-x86 + +# Add ICC (from /lrde/dev) to the PATH. +path_prepend intel/icc9.1 + # The ccache dir is located in the /work file system. As ~/.zshenv is # not sourced by the build script, this environment variable is # defined here. @@ -17,13 +22,10 @@ # Initialize ICC environment variables. init_icc_vars() { - # ICC 9.0 directory. - icc_dir="/lrde/beyrouth/admin/lrde/usr/intel/compiler90" - - # Help icpc. - export CPPFLAGS="-D_LINUX -I$icc_dir/include/c++/" + # ICC 9.1 settings. + source /lrde/dev/linux-x86/intel/icc9.1/bin/iccvars.sh - # BTW, disable some (sometimes fallacious) remarks and warnings from icpc. + # Disable some (sometimes fallacious) remarks and warnings from icpc. # # FIXME: Some of these warnings could be disabled from configure when # the selected C++ compiler is icpc. @@ -202,27 +204,7 @@ # ignored_remarks="279,304,383,424,444,593,810,869,1292,1418,1572,1682,1683" ignored_warnings="61,68,161,239,592,854,858,1292" - export CXXFLAGS="-wd$ignored_remarks,$ignored_warnings" - export LDFLAGS="-cxxlib-icc -i-static" - export LD_LIBRARY_PATH="$icc_dir/lib:`printenv LD_LIBRARY_PATH`" - - # The directory /work/build/bin-icc contains replacement wrappers for - # icc and icpc, because the installation of ICC on the lrde account - # is broken since the reorganization of the accounts (the lrde - # account has been moved from /home/lrde/admin/lrde to - # /lrde/beyrouth/admin/lrde). - # - # - # Moreover, ICC 9.0 cannot work with GCC 4.1 installed on the - # same host. icpc reports this error: - # - # Command-line error: invalid GNU version number: 412 - # - # Yet GCC 4.1 is the default version of GCC on brasilia. So, - # /work/build/bin-icc also contains symlinks for gcc and g++ - # pointing to the version 4.0 of the GNU C and C++ compilers. - export PATH="/work/build/bin-icc:$icc_dir/bin:$PATH" } @@ -258,8 +240,8 @@ ) ( - export CC="ccache icc" - export CXX="ccache icpc" + export CC="icc" + export CXX="icpc" init_icc_vars test_tree vaucanson-icc \ bootstrap configure build check demos install distcheck @@ -270,8 +252,8 @@ # Olena. # # ------- # -# TEST_OLENA_TREE TREE CXX-COMPILER -# --------------------------------- +# TEST_OLENA_TREE TREE CXX-COMPILER [OPTIONS]... +# ---------------------------------------------- # A wrapper of test_tree() for Olena trees. TREE is the name of the # project (either olena-proto-1.0 or olena-1.0), and CXX_COMPILER is # the C++ compiler used. @@ -280,58 +262,37 @@ # Work inside a sub-shell to avoid environment variables escaping. ( tree=$1 - cxx_compiler=$2 - # Use ccache. - export CXX="ccache $cxx_compiler" + shift + cxx_compiler=$1 + shift # ICC-specific initializations, if needed. - if test "x$cxx_compiler" = xicpc; then + case "$cxx_compiler" in + icpc*) init_icc_vars - fi - test_tree "$tree-$cxx_compiler" + # Do not use ccache with icpc, as icpc tends to mix up with + # generated dependencies. + export CXX="$cxx_compiler" + ;; + *) + # Use ccache. + export CXX="ccache $cxx_compiler" + ;; + esac + test_tree "$tree-$cxx_compiler" "$@" ) } # Olena 0.10a. - -# FIXME: 2006-09-18: Disabled, since it ate too much CPU ressources on -# brasilia (in fact, Olena 0.10a builds cycles were endless!). -# So we're back to the manual updates for Olena 0.10a. - - -# # ACTION_FETCH_OLN_FROM_PRCS -# # -------------------------- -# # An additional action to fetch Olena O.10a from beyrouth's PRCS -# # repository. -# # -# # Any tested source tree is normally populated on goa, and fetched from -# # this machine; but the PRCS repository, where Olena 0.10a lies, is not -# # accessible from goa -- but brasilia can! Hence this function, -# # triggered as an action by test_tree() before the boostraping of the -# # source tree. -# action_fetch_oln_from_prcs() -# { -# ( -# export PRCS_REPOSITORY=~prcs/repository -# # The --force option is required to run prcs with no terminal. -# prcs checkout --force oln -# ) -# } - -# # FIXME: 2006-09-16: As the working copy of Olena 0.10a is always -# # empty on goa, we need to always force the build by removing its -# # checksum. -# rm -f build.olena-0.10a.sum -# # FIXME: 2006-09-16: Augment maintainer/update_unpacked with the -# # ability to fetch copies from the PRCS repository instead of using -# # the dedicated action `fetch_oln_from_prcs'. -# test_tree olena-0.10a \ -# fetch_oln_from_prcs bootstrap configure build check install distcheck - -test_tree olena-0.10a bootstrap configure build check install distcheck - +# ------------ +# FIXME: Augment `maintainer/update_unpacked' with the ability to fetch +# copies from the PRCS repository instead of using manual updates from +# brasilia using `maintainer/update_unpacked-oln-0.10a.'? +test_olena_tree olena-0.10a g++-4.1 +test_olena_tree olena-0.10a icpc-9.1 # Olena proto-1.0. +# ---------------- # 2006-09-16: Disabled. # test_olena_tree olena-proto-1.0 g++-3.4 # test_olena_tree olena-proto-1.0 g++-4.0 @@ -339,8 +300,8 @@ # test_olena_tree olena-proto-1.0 icpc # Olena 1.0. +# ---------- test_olena_tree olena-1.0 g++-3.4 test_olena_tree olena-1.0 g++-4.0 test_olena_tree olena-1.0 g++-4.1 - -test_olena_tree olena-1.0 icpc +test_olena_tree olena-1.0 icpc-9.1

1 0

We bought the house in 1997, back when I was still working full time at a programming job.
by Rosamond Swanson 12 Feb '07

12 Feb '07

1 0

Scientists have been revising how common they think the disorder is.
by Baldwin V. Costello 10 Feb '07

10 Feb '07

According to UNAIDS, 16. Chavez said the traditional American diet is high in fat, "not just Mexican food but every good food. "It is the little pieces of the puzzle that will eventually put together a holistically successful presidency. "Authorities are planning to open another sex museum in anorthern suburb, away from the red light district, to attract awider audience. "The patients are in good general shape. While HIV is neither as fast a killer nor as contagious as influenza, it is equally large a threat. Botswana, Zambia and Lesotho are likely to be most affected by the funding gap. "We however, found that those interviewed, like all Kenyans, are rather reserved about their sex life. "COMBI is not only about producing posters and T-Shirts. On Thursday, parents and religious leaders reacted with shock and disgust at these findings, saying something needs to be done urgently to address society's rapidly deteriorating moral standards. Managed Ethernet Switches withstand harsh environments. There are an estimated 5. "South Africa is an excellent choice for this trial. 'Natural' Remedies May Interfere With Clinical Trials - Yahoo!

1 0

[LrdeTools] 416: Repair svn-stats.
by Roland Levillain 09 Feb '07

09 Feb '07

https://svn.lrde.epita.fr/svn/lrde-tools/trunk/build-farm Index: ChangeLog from Roland Levillain <roland(a)lrde.epita.fr> Repair svn-stats. * svn-stats/svn-stats: Temporarily disable TC, since we have no privileged access to the repository as of now. (commits, commits_per_contributor): Pass the year as an additional argument. Adjust callers. * svn-stats/README: Catch up with the move of svn-stats from marvejols to brasilia. README | 3 ++- svn-stats | 23 ++++++++++++----------- 2 files changed, 14 insertions(+), 12 deletions(-) Index: svn-stats/svn-stats --- svn-stats/svn-stats (revision 415) +++ svn-stats/svn-stats (working copy) @@ -37,7 +37,7 @@ "Olena 1.0 (Metalic)" => 'https://svn.lrde.epita.fr/svn/oln/trunk/metalic', "Olena 1.0 (Extended)" => 'https://svn.lrde.epita.fr/svn/oln/trunk/extended', "Olena 1.0 (Static)" => 'https://svn.lrde.epita.fr/svn/oln/trunk/static', - TC => 'https://svn.lrde.epita.fr/svn/tc', +# TC => 'https://svn.lrde.epita.fr/svn/tc', Havm => 'https://svn.lrde.epita.fr/svn/havm', Nolimips => 'https://svn.lrde.epita.fr/svn/nolimips', MonoBurg => 'https://svn.lrde.epita.fr/svn/monoburg', @@ -84,7 +84,7 @@ foreach (sort keys %proj) { print STDOUT "$_\n"; - &commits($_, $proj{$_}); + &commits($_, $proj{$_}, 2007); } my $date = localtime; @@ -98,7 +98,7 @@ close REPORT; sub commits { - my ($name, $repository) = @_; + my ($name, $repository, $year) = @_; my @commits_week; my %commits; my $revision; @@ -116,7 +116,8 @@ $name = $alias{$name} if defined $alias{$name}; $commits{$name}++; my $time = timegm(0, 0, 0, $5, $4 - 1, $3); - $commits_week[&week_from_time($time) - 1]++ if $3 eq '2006'; + $commits_week[&week_from_time($time) - 1]++ + if $3 eq $year; } print STDOUT "\n"; @@ -130,7 +131,7 @@ $max += 49; $max -= $max % 50; - &commits_per_contributor(\@data, "$report_dir/$name.1.png", $max); + &commits_per_contributor(\@data, "$report_dir/$name.$year.1.png", $max); # @data = (); # foreach (sort { $commits{$b} <=> $commits{$a} } keys %commits) { @@ -141,7 +142,7 @@ # use Data::Dumper; # print Dumper(\@data); # -# &commits_per_contributor_per_week(\@data, "$report_dir/$name.3.png"); +# &commits_per_contributor_per_week(\@data, "$report_dir/$name.$year.3.png"); my $i = 1; @data = (); @@ -159,13 +160,13 @@ $max += 9; $max -= $max % 10; - &commits_per_week(\@data, "$report_dir/$name.2.png", $max); + &commits_per_week(\@data, "$report_dir/$name.$year.2.png", $max, $year); print <<HTML; <h2>$name (<span style="color:green">$revision</span>)</h2> <div> - <img src="$name.1.png" alt="$name 1" /> - <img src="$name.2.png" alt="$name 2" /> + <img src="$name.$year.1.png" alt="$name 1" /> + <img src="$name.$year.2.png" alt="$name 2" /> </div> <hr/> HTML @@ -197,12 +198,12 @@ } sub commits_per_week { - my ($data, $filename, $max) = @_; + my ($data, $filename, $max, $year) = @_; my $graph = GD::Graph::lines->new(600, 300); $graph->set( y_label => 'Number of commits', - x_label => 'Week in 2006', + x_label => 'Week in ' . $year, title => 'Commits per week', dclrs => [ 'red' ], valuesclr => 'black', Index: svn-stats/README --- svn-stats/README (revision 415) +++ svn-stats/README (working copy) @@ -4,6 +4,7 @@ statistics on commits to the Subversion repository of the LRDE. svn-stats is to be launched regularly, through cron; for instance, the -crontab of the user `doc' on host `marjevols' contains: +crontab of the user `doc' on host `brasilia' contains: +# Calcul des statistiques sur le dépôt Subversion. 0 2 * * * /work/build/svn-stats/svn-stats && rsync -e "ssh -i /work/build/build_farm/id_dsa_build -l pm" -qr /tmp/stats/* goa::svn-stats/

1 0

Vcs: Send messages in Latin-1 by default
by Roland Levillain 09 Feb '07

09 Feb '07

Hello, Voici deux propositions de patches pour Vcs et RubyEx qui permettent d'indiquer le jeu de caractères et l'encodage choisi pour poster des mail avec Vcs. Les deux sont par faits rapport à Vcs 0.5 beta 10 Le premier patch concerne Vcs, le second RubyEx, localisé dans vendor/ruby_ex dans la distribution de Vcs. Comment on fait pour soumettre ces patches et éventuellement les enregistrer ? 2007-02-09 Roland Levillain <roland(a)lrde.epita.fr> * lib/vcs/mail.rb (Vcs::default_options): Add two fields, `charset' and `encoding', to send mail in Latin-1 by default in Vcs. --- lib/vcs/mail.rb.orig 2007-02-09 11:18:32.000000000 +0100 +++ lib/vcs/mail.rb 2007-02-09 11:19:04.000000000 +0100 @@ -12,6 +12,9 @@ @@mailer ||= Sendmail.new @@default_options ||= { :mime => true, + # Use Latin-1 by default. + :charset => 'iso-8859-1', + :encoding => 'quoted-printable', :header => { 'X-Mailer' => "Vcs.mail (#{Vcs::Version})" } } cattr_accessor :mail_options 2007-02-09 Roland Levillain <roland(a)lrde.epita.fr> * lib/sendmail.rb (Sendmail::parse_mail_options): Add two fields in the options, `charset' and `encoding', selecting plain 7-bit ASCII by default, that the user can override. Add two options corresponding to these fields. Typos. --- vendor/ruby_ex/lib/sendmail.rb.orig 2007-02-09 11:18:12.000000000 +0100 +++ vendor/ruby_ex/lib/sendmail.rb 2007-02-09 11:48:43.000000000 +0100 @@ -20,7 +20,9 @@ :to => [], :server => ENV['SMTPSERVER'] || 'localhost:25', :header => {}, - :body => STDIN + :body => STDIN, + :charset => 'us-ascii', + :encoding => '7bit' ) string_args, hash_args = args.partition { |x| x.is_a?(String) } hash_args.each do |hash| @@ -75,9 +77,17 @@ opts.on('--header STRING', 'Add some header fields (Yaml syntax)') do |s| options.header.merge! YAML.load(s) end - opts.on('-m', '--[no-]mime', 'Choose the mime protocole') do |mime| + opts.on('-m', '--[no-]mime', 'Choose the MIME protocol') do |mime| options.mime = mime end + opts.on('--charset STRING', + 'Choose a charset for the encoding') do |aCharset| + options.charset = aCharset + end + opts.on('--encoding STRING', + 'Choose an encoding') do |anEncoding| + options.encoding = anEncoding + end opts.on_tail('-h', '--help', 'Show this message') do puts opts exit @@ -156,11 +166,11 @@ TempPath.new do |mail_body| mail_body.open('w') do |out| if opts.mime - out.puts ' - |Content-Type: text/plain; charset="us-ascii" - |Content-Transfer-Encoding: 7bit + out.puts " + |Content-Type: text/plain; charset=\"#{opts.charset}\" + |Content-Transfer-Encoding: #{opts.encoding} | - |'.head_cut! + |".head_cut! end out.print body end

1 0

[LrdeTools] 415: Remove tanis.
by Roland Levillain 09 Feb '07

09 Feb '07

https://svn.lrde.epita.fr/svn/lrde-tools/trunk/build-farm Tanis is no longer used (it is warming up very quickly, and is not usable as a host for the build farm). BTW, a small stress test for Vcs: á à â ä ã é è ê ë í ì î ï ó ò ô ö õ ú ù û ü ý ÿ ç Á À Â Ä Ã É È Ê Ë Í Ì Î Ï Ó Ò Ô Ö Õ Ú Ù Û Ü Ý Ç ¹ ² ³ ¡ ¿ ° © :P Index: ChangeLog from Roland Levillain <roland(a)lrde.epita.fr> Remove tanis. * buildfarm_worker/tanis.fns: Remove. 0 files changed

1 0

Re: argp r5: Fix the check for `restrict'.
by Roland Levillain 09 Feb '07

09 Feb '07

SIGOURE Benoit <tsuna(a)lrde.epita.fr> writes: > URL: https://svn.lrde.epita.fr/svn/argp/trunk > > ChangeLog: > 2007-02-02 SIGOURE Benoit <tsuna(a)lrde.epita.fr> > > Fix the check for `restrict'. > * argp-fmtstream.h [!HAVE_UNISTD_H]: Don't use #define to change > ssize_t in int, use a typedef instead. > * argp.h [!__restrict]: Fix the conditional definition of __restrict > (remove a typo, use C99 features). [...] Thanks for synchronizing with the original argp and your last two patches! BTW, did you warn the upstream author that the build system was using obsolete (and broken) constructs?

3 5

You Are My Guiding Star
by Webster 08 Feb '07

08 Feb '07

1 0

Security: We have limited access to your PayPaI account
by PayPaI Security 07 Feb '07

07 Feb '07

1 0

argp r5: Fix the check for `restrict'.
by SIGOURE Benoit 02 Feb '07

02 Feb '07

URL: https://svn.lrde.epita.fr/svn/argp/trunk ChangeLog: 2007-02-02 SIGOURE Benoit <tsuna(a)lrde.epita.fr> Fix the check for `restrict'. * argp-fmtstream.h [!HAVE_UNISTD_H]: Don't use #define to change ssize_t in int, use a typedef instead. * argp.h [!__restrict]: Fix the conditional definition of __restrict (remove a typo, use C99 features). argp-fmtstream.h | 2 +- argp.h | 11 ++++++----- 2 files changed, 7 insertions(+), 6 deletions(-) Index: trunk/argp.h =================================================================== --- trunk/argp.h (revision 4) +++ trunk/argp.h (revision 5) @@ -43,14 +43,15 @@ # define __error_t_defined #endif -/* FIXME: What's the right way to check for __restrict? Sun's cc seems - not to have it. Perhaps it's easiest to just delete the use of - __restrict from the prototypes. */ #ifndef __restrict -# ifndef __GNUC___ +# if defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L +# define __restrict restrict +# else /* No C99: */ +# ifndef __GNUC__ # define __restrict # endif -#endif +# endif /* C99 */ +#endif /* __restrict */ /* NOTE: We can't use the autoconf tests, since this is supposed to be an installed header file and argp's config.h is of course not Index: trunk/argp-fmtstream.h =================================================================== --- trunk/argp-fmtstream.h (revision 4) +++ trunk/argp-fmtstream.h (revision 5) @@ -35,7 +35,7 @@ /* This is a kludge to make the code compile on windows. Perhaps it would be better to just replace ssize_t with int through out the code. */ -# define ssize_t int +typedef ssize_t int; #endif #if _LIBC || (defined (HAVE_FLOCKFILE) && defined(HAVE_PUTC_UNLOCKED) \ -- SIGOURE Benoit aka Tsuna (SUSv3 compliant) _____ "On a long enough timeline, the survival rate /EPITA\ Promo 2008.CSI/ACU/YAKA for everyone drops to zero" -- Jack.

1 0