Monthly Archives: January 2007

Stopped smoking

Cigarette Ashtray
While reading this entry. I just remembered that it has been a a litle bit more than a month that i stopped smoking. I was a real smoker, smoking a 25 red malboro every day. But one day i wake up (one of those
days you know) and like that i thought to maybe stop smoking. So far it has been good, the hardset part is the first days really, after it get OK. I hope it will last for me, i feel so much better i think since i stopped smoking.

Emacs nighly cvs snapshot with xft on Ubuntu Edgy

I wanted to try the latest cvs snapshot with XFT support, since i did not want to screw up more my workstation i have used packages instead of make install blindy.

Basically i have a script called ./

set -e

d=$(date '+%Y%m%d')

mkdir -p cvs

pushd cvs >/dev/null && {
cvs -Q -z3 co -r emacs-unicode-2 emacs
} && popd >/dev/null

mkdir -p build
[[ -d build/emacs-${d} ]] && rm -rf build/emacs-${d}
cp -al cvs/emacs build/emacs-${d}

zcat patches/emacs-snapshot_${debpatch}.diff.gz|patch -p1 -d build/emacs-${d}
cat patches/with-font.patch|patch --silent -p1 -d build/emacs-${d}

pushd build/emacs-${d} >/dev/null && {
    chmod +x debian/rules
    dch -v "1:${d}-1" "New snapshot."
    dch "Build with xft."
    fakeroot dpkg-buildpackage -b
} && popd >/dev/null

in patches/with-font.patch i have :

--- c/debian/rules.chmou        2007-01-21 23:21:09.486353750 +1100
+++ c/debian/rules      2007-01-21 23:21:13.914630500 +1100
@@ -393,7 +393,7 @@
 # Emacs-gtk confflags
 emacs_gtk_confflags := ${confflags}
 emacs_gtk_confflags += --with-x=yes
-emacs_gtk_confflags += --with-x-toolkit=gtk
+emacs_gtk_confflags += --with-x-toolkit=gtk  --enable-font-backend --with-xft

 # Emacs-nox confflags
 emacs_nox_confflags := ${confflags}
--- c/src/emacs.c.chmou 2007-01-21 23:21:09.486353750 +1100
+++ c/src/emacs.c       2007-01-21 23:22:18.430662500 +1100
@@ -1408,10 +1408,10 @@
     = argmatch (argv, argc, "-nl", "--no-loadup", 6, NULL, &skip_args);

-  enable_font_backend = 0;
-  if (argmatch (argv, argc, "-enable-font-backend", "--enable-font-backend",
-               4, NULL, &skip_args))
     enable_font_backend = 1;
+  if (argmatch (argv, argc, "-disable-font-backend", "--disable-font-backend",
+               4, NULL, &skip_args))
+    enable_font_backend = 0;
 #endif /* USE_FONT_BACKEND */

@@ -1816,7 +1816,7 @@
   { "-unibyte", "--unibyte", 81, 0 },
   { "-no-multibyte", "--no-multibyte", 80, 0 },
   { "-nl", "--no-loadup", 70, 0 },
-  { "-enable-font-backend", "--enable-font-backend", 65, 0 },
+  { "-disable-font-backend", "--disable-font-backend", 65, 0 },
   /* -d must come last before the options handled in startup.el.  */
   { "-d", "--display", 60, 1 },
   { "-display", 0, 60, 1 },

i have as well in patches/ the ubuntu (or could be debian) patch downloaded from the ubuntu (or debian archive) archive which is for ubuntu on

If there is a new version you will need to increase the version in to match the patch downloaded.

When running it should produce binary in build/* with xft enabled by default. Make sure to have all the dependencies (dpkg-buildpackages should tell you if there is unresovled one).

One screenshot :

Screenshot of Emacs snapshot with XFT

History expansion and substitions in ZSH

I better to keep that somewhere since i always forget that thing, to do a search and replace from the command line in zsh. you just have to do the :s^FOO^BAR after your expansion

For example you just have typed the long command line :

blah bar FOO=1 FOO=3 FOO=6 cnt=1

you can just type :


and it will be expanded to :

blah bar VALUE=1 VALUE=3 VALUE=6 cnt=1

Benchmarking with PGBench: TPS results.

pgbench is not the best tools to benchmark postgres but it’s
the one shipped by default with postgres.
It does not do a good job for benchmarking your “web apps” SQL wise, it will give you some
good indication about how fast is your server for postgres.

Here are some results we have collected using theses options :

pgbench -Uuser  -s 10 -c 10 -t 3000 benchmark

We configure servers with RAID 1+0 with write-cache and a battery
backup, we have tested with postgres 8.1.5 and here are the best
results for different type of servers :

IBM x3650 : 729tps
HP DL385 : 717tps
Dell PowerEdge 2950: 708tps