Chromium browser

**Grogan** · 08-27-2011, 01:50 AM

Well, I finally got the gumption to try this on Linux. I refuse to use anything I can't compile myself, so that was a requirement. (You can't really compile "Google Chrome" itself, it's proprietary). Chromium is the open source portion, and it's now usable. (It has no install, but that's easily done manually because it's a big fat binary result (50 megs here, 64 bit), and just a few libraries and resources. It can pretty much run out of a directory. I put mine in /opt/chromium, stripped the binaries and made some symlinks. Distributors are packaging this up nowadays.

Anyway, it was a motherfuckin futhermucker, but I have a viable build procedure now for this. I'm going to see if I can use it. (So far, so good... the flash plugin works and shit.). The problem was, all the dependencies. I found switches to lose most of them (PAM, Gnome rubbish etc.) so I didn't have to junk up my system.

Holy Snappin Arseholes... is this browser ever fast on Linux (compiled for the CPU and without being linked to a bunch of bloat). I have not seen the like.

There aren't many options, but at least I can get it to use my black GTK+ style instead of the default Google Chrome look that I hate. I can also get it to use my window manager's window manager borders and window controls, but that's kind of redundant.

This fits in to my black desktop style rather nicely:

Go ahead and piss me off some more, Firefox. Make your browser any more like Google Chrome, and I might as well use... this

**Grogan** · 08-27-2011, 04:58 AM

Hardware Acceleration works too (have to enable it in about:flags)

I get the full "60+ FPS" in this test (25 without it enabled)

http://demos.hacks.mozilla.org/openweb/HWACCEL/

Also, things like big long slashdot threads render instantly, and scroll effortlessly.

A few quirks with Vbulletin 4 though, like if I want to edit a post I have to click "Edit Post" twice (it doesn't complete the first time... just stays with busy cursor, but clicking it again brings up the editor in advanced mode)

**Bad Haircut** · 08-27-2011, 05:25 AM

If it ticks all the boxes go for it, be interesting to see if you can live with it long term and wins you over.

**mlangdn** · 08-27-2011, 07:59 AM

Running FF-7.0b1 I got 38 FPS. Interesting test! FF-7 seems to render much faster also.

**MSUredux** · 08-27-2011, 08:15 AM

I get 60+ FPS with IE 9 in Win7 64bit

**Bad Haircut** · 08-27-2011, 09:06 AM

^ Same here on Firefux

**mlangdn** · 08-27-2011, 09:34 AM

Ok, so my curiosity is peaked - Chromium is compiling right now. That tarball was 139 megs. No wonder it has a big, fat binary.

**ZemaTalon** · 08-27-2011, 12:12 PM

I've been using Chromium and Chrome, along side FF for some time now, though I did it the quick and dirty way and installed Chromium via the Ubuntu repositories

EDIT: Forgot to add Seamonkey to that list, as of about a week ago. FF7 beta has been running great on my other machine so I just upgraded my main system over to it. Now to see if I can abuse it longer with a zillion open tabs

**mlangdn** · 08-27-2011, 01:19 PM

Chromium is up and running. That test from above really spins like crazy for 60+! That said, I don't see it as any faster than FF-7.
Well, hold on a sec - I just loaded Fox News very daggone quickly. Guess I need to use this a bit more to be fair.

**Grogan** · 08-27-2011, 01:33 PM

Originally Posted by mlangdn

Running FF-7.0b1 I got 38 FPS. Interesting test! FF-7 seems to render much faster also.

Under Linux, you may have to force enable hardware acceleration with an environment variable: MOZ_GLX_IGNORE_BLACKLIST=1 unless using Xorg DRM I think. (e.g. start Firefox with "MOZ_GLX_IGNORE_BLACKLIST=1 firefox" in a terminal)

Here's why they don't want to just enable it arbitrarily:

https://cvs.khronos.org/svn/repos/re...nce-tests.html

You'll be lucky if your browsers don't crash on some of those tests.

It's much easier for Internet Explorer to do this, because it doesn't care about being portable. Microsoft just uses DirectX and lets it decide whether you can have hardware acceleration or not. Microsoft has DELIBERATELY obstructed OpenGL over the years (because it's an open system that competes with their proprietary APIs), which has greatly harmed its adoption, and makes the video card vendors support it directly. They even REMOVE OpenGL support from the drivers they bundle. Fuck them.

Firefox in Windows also uses DirectX, so hardware acceleration is enabled by default in Windows. This means that they have to work twice, instead of using portable APIs.

**MSUredux** · 08-27-2011, 01:45 PM

Originally Posted by Grogan

Microsoft has DELIBERATELY obstructed OpenGL over the years

Yeah, that kind of pisses me off. I have several older games that used OpenGL and I had to go find modified executables that would run with DirectX. It was either that or play them with crappy graphics.

**mlangdn** · 08-27-2011, 01:46 PM

Results: (7003 of 7124 passed)
From FF-7

Chromium crashed 5 seconds in.

I'm gonna try the test again with "MOZ_GLX_IGNORE_BLACKLIST=1 firefox"

In a terminal started FF now, and got the same result. Which is mostly Greek to me anyway.

**Grogan** · 08-27-2011, 02:03 PM

My Chromium (15.0.860.0) passed most of those tests, but crashed near the end and I never got to see the final result. Correction, the tab processes crashed, not the parent. I just had to click the reload button on the other tabs (didn't even lose my text while composing a post). That is unusual (usually it would just be one tab process that says "Ohh... Snap!") but this is a pretty tortuous test, and uses direct rendering. All bets are off trying to control that. Even the kernel can't survive a video driver crash sometimes.

That's actually another point in favour of chome... it's worth the wasted memory to have every tab a separate process.

P.S. In a few days I'm going to try a newer tarball. I had to backtrack two days because some of the python scripts are currently broken or out of sync. (Forget SVN... you'll never complete the steps to generate the project files unless you HAVE all the prerequisites, even though you plan on disabling them for the build in a subsequent configuration step with Gyp. I wasted a whole night fucking with that. I am now getting tarballs from http://build.chromium.org/official/ instead)

**Grogan** · 08-27-2011, 03:28 PM

Originally Posted by mlangdn

No wonder it has a big, fat binary.

Well, it's a big fat binary because there is a lot of static linkage and functionality built right in instead of abstracted out into libraries. Firefox is actually bigger when you consider the sum of all the parts.

However, it's of significance to note that Chromium actually USES mozilla libraries (nspr, nss etc.)

Look at this... because this is where I keep my netscape runtimes, Chromium is actually linked to libraries in my Firefox directory (I do have to properly install nspr and nss for these Chromium builds, I built a package specifically for that, but I remove it after because I want the linker to actually go for the Firefox libraries. Firefox is the most important user of those libraries and so they should come from there instead of possibly stagnating libraries installed to the system. I use ld.so.conf entries so the mozilla libraries will be found there.)

Code:

$ ldd chrome
        linux-vdso.so.1 =>  (0x00007fff20bff000)
        libX11.so.6 => /usr/lib64/libX11.so.6 (0x00007f46ecaba000)
        libdl.so.2 => /lib64/libdl.so.2 (0x00007f46ec8b6000)
        libXrender.so.1 => /usr/lib64/libXrender.so.1 (0x00007f46ec6ac000)
        libXss.so.1 => /usr/lib64/libXss.so.1 (0x00007f46ec4a9000)
        libXext.so.6 => /usr/lib64/libXext.so.6 (0x00007f46ec297000)
        librt.so.1 => /lib64/librt.so.1 (0x00007f46ec08e000)
        libgtk-x11-2.0.so.0 => /usr/lib64/libgtk-x11-2.0.so.0 (0x00007f46eba58000)
        libgdk-x11-2.0.so.0 => /usr/lib64/libgdk-x11-2.0.so.0 (0x00007f46eb7a6000)
        libpangocairo-1.0.so.0 => /usr/lib64/libpangocairo-1.0.so.0 (0x00007f46eb59a000)
        libgdk_pixbuf-2.0.so.0 => /usr/lib64/libgdk_pixbuf-2.0.so.0 (0x00007f46eb37c000)
        libcairo.so.2 => /usr/lib64/libcairo.so.2 (0x00007f46eb0cf000)
        libpango-1.0.so.0 => /usr/lib64/libpango-1.0.so.0 (0x00007f46eae84000)
        libfreetype.so.6 => /usr/lib64/libfreetype.so.6 (0x00007f46eabfd000)
        libfontconfig.so.1 => /usr/lib64/libfontconfig.so.1 (0x00007f46ea9c8000)
        libgobject-2.0.so.0 => /usr/lib64/libgobject-2.0.so.0 (0x00007f46ea779000)
        libgthread-2.0.so.0 => /usr/lib64/libgthread-2.0.so.0 (0x00007f46ea575000)
        libglib-2.0.so.0 => /usr/lib64/libglib-2.0.so.0 (0x00007f46ea26f000)
        libevent-2.0.so.5 => /usr/local/lib/libevent-2.0.so.5 (0x00007f46ea02c000)
        libnss3.so => /opt/firefox/lib/firefox-6.0/libnss3.so (0x00007f46e9d22000)
        libsmime3.so => /opt/firefox/lib/firefox-6.0/libsmime3.so (0x00007f46e9afa000)
        libnssutil3.so => /opt/firefox/lib/firefox-6.0/libnssutil3.so (0x00007f46e98de000)
        libplc4.so => /opt/firefox/lib/firefox-6.0/libplc4.so (0x00007f46e96da000)
        libnspr4.so => /opt/firefox/lib/firefox-6.0/libnspr4.so (0x00007f46e949c000)
        libz.so.1 => /usr/lib64/libz.so.1 (0x00007f46e9288000)
        libresolv.so.2 => /lib64/libresolv.so.2 (0x00007f46e906d000)
        libjpeg.so.8 => /usr/lib64/libjpeg.so.8 (0x00007f46e8e35000)
        libxml2.so.2 => /usr/lib64/libxml2.so.2 (0x00007f46e8ad8000)
        libasound.so.2 => /usr/lib64/libasound.so.2 (0x00007f46e87f6000)
        libxslt.so.1 => /usr/lib64/libxslt.so.1 (0x00007f46e85ba000)
        libcups.so.2 => /usr/lib64/libcups.so.2 (0x00007f46e837a000)
        libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f46e815d000)
        libgcrypt.so.11 => /lib64/libgcrypt.so.11 (0x00007f46e7ee6000)
        libbz2.so.1 => /lib64/libbz2.so.1 (0x00007f46e7cd6000)
        libdbus-glib-1.so.2 => /usr/lib64/libdbus-glib-1.so.2 (0x00007f46e7ab2000)
        libdbus-1.so.3 => /usr/lib64/libdbus-1.so.3 (0x00007f46e786e000)
        libexpat.so.1 => /usr/lib64/libexpat.so.1 (0x00007f46e764b000)
        libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00007f46e7347000)
        libm.so.6 => /lib64/libm.so.6 (0x00007f46e70c3000)
        libgcc_s.so.1 => /usr/lib64/libgcc_s.so.1 (0x00007f46e6eae000)
        libc.so.6 => /lib64/libc.so.6 (0x00007f46e6b0a000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f46ecdf8000)
        libxcb.so.1 => /usr/lib64/libxcb.so.1 (0x00007f46e68ee000)
        libXau.so.6 => /usr/lib64/libXau.so.6 (0x00007f46e66eb000)
        libXdmcp.so.6 => /usr/lib64/libXdmcp.so.6 (0x00007f46e64e6000)
        libXinerama.so.1 => /usr/lib64/libXinerama.so.1 (0x00007f46e62e4000)
        libXi.so.6 => /usr/lib64/libXi.so.6 (0x00007f46e60d5000)
        libXrandr.so.2 => /usr/lib64/libXrandr.so.2 (0x00007f46e5ecd000)
        libXcursor.so.1 => /usr/lib64/libXcursor.so.1 (0x00007f46e5cc3000)
        libXcomposite.so.1 => /usr/lib64/libXcomposite.so.1 (0x00007f46e5ac1000)
        libXdamage.so.1 => /usr/lib64/libXdamage.so.1 (0x00007f46e58bf000)
        libXfixes.so.3 => /usr/lib64/libXfixes.so.3 (0x00007f46e56ba000)
        libatk-1.0.so.0 => /usr/lib64/libatk-1.0.so.0 (0x00007f46e5499000)
        libpixman-1.so.0 => /usr/lib64/libpixman-1.so.0 (0x00007f46e522c000)
        libpng14.so.14 => /usr/lib64/libpng14.so.14 (0x00007f46e5007000)
        libgio-2.0.so.0 => /usr/lib64/libgio-2.0.so.0 (0x00007f46e4cf6000)
        libpangoft2-1.0.so.0 => /usr/lib64/libpangoft2-1.0.so.0 (0x00007f46e4ac9000)
        libgmodule-2.0.so.0 => /usr/lib64/libgmodule-2.0.so.0 (0x00007f46e48c6000)
        libplds4.so => /opt/firefox/lib/firefox-6.0/libplds4.so (0x00007f46e46c3000)
        libssl.so.0 => /lib64/libssl.so.0 (0x00007f46e4471000)
        libcrypto.so.0 => /lib64/libcrypto.so.0 (0x00007f46e40e8000)
        libcrypt.so.1 => /lib64/libcrypt.so.1 (0x00007f46e3eaf000)
        libgpg-error.so.0 => /usr/lib64/libgpg-error.so.0 (0x00007f46e3cac000)

**Dark Angel** · 11-01-2011, 06:41 AM

Got jack of the shitty page loading in FF4 (Slack 13.37 default) and compiled Chromium to see how it'd go.

FF4 scored about 9FPS on my system
Chromium per defaults scored 19FPS
Enabling GPU functions punched that to 60+

dayum!

**mlangdn** · 11-01-2011, 07:09 AM

Here's a Chromium alt that I'm giving a spin:

http://www.srware.net/en/software_srware_iron.php

SRWare Iron: The browser of the future - based on the free Sourcecode "Chromium" - without any problems at privacy and security

Google's Web browser Chrome thrilled with an extremely fast site rendering, a sleek design and innovative features. But it also gets critic from data protection specialists , for reasons such as creating a unique user ID or the submission of entries to Google to generate suggestions. SRWare Iron is a real alternative. The browser is based on the Chromium-source and offers the same features as Chrome - but without the critical points that the privacy concern.

I'm using their binary, rather than compiling from source, for a quick test drive. I had to install gconf and orbit-2. This baby is really fast - especially with the adblock extension installed.

**Grogan** · 11-01-2011, 12:27 PM

You should not have installed gconf and orbit-2 (gnome garbage) because now other things you compile (probably anything gtk+ based) are going to link against those and they are going to be dependencies. You may not be able to configure around them if they are present. (e.g. don't expect a --disable for them. That has to be specifically written into the configure scripts and macros)

So be sure and remove those when done playing with that binary. (you can also delete their includes and pkgconfig files and keep the libraries for runtime only)

**mlangdn** · 11-01-2011, 02:45 PM

I probably won't keep it long. I haven't fired up chromium for a while either. It seems I'm a FF junkie!

**Dark Angel** · 11-01-2011, 08:15 PM

Well afyer saying how fast it is, and it was damn fast, I've removed it. For some reason it wouldn't show the new/top posts thing when I clicked Forum here and it made posts at another forum almost unreadable. When I get home after my exams I'll try with the latest source but I'm suspicious it's an engine issue. Happy to be wrong though. I'd also like to try compiling the latest Firefox as a comparison.

**ZemaTalon** · 11-01-2011, 08:51 PM

Originally Posted by Dark Angel

Well afyer saying how fast it is, and it was damn fast, I've removed it. For some reason it wouldn't show the new/top posts thing when I clicked Forum here and it made posts at another forum almost unreadable. When I get home after my exams I'll try with the latest source but I'm suspicious it's an engine issue. Happy to be wrong though. I'd also like to try compiling the latest Firefox as a comparison.

Sounds similar to the problem me and several others here were having with FF 3.6x. Me and Gro have been using Chromium here on the forum for a while now. I haven't had any major issues, other than an occasional tendency to cause a double post, which you've probably noticed already

**Grogan** · 11-01-2011, 09:55 PM

The double post bug is gone as well and it has been for a long time. Those are the mistakes I made, thinking that my problems were the same old ones I was reading about in Google searches, that weren't fixed. I also made the mistake of thinking it was engine related. Mine was miscompiled... the browser was not working correctly with my version of libxml. The posting bug and the AJAX issues were symptoms of the same problem, for me.

Versions in any distro repository are going to be very much out of date. If you are still having the double post bug, it is likely to be because you're still using an old version.

My builds have been working perfectly now. I don't think I have any issues whatsoever at any sites I go to. A lot of the problems with Flash content were fixed by... upgrading my mozilla flash plugin to the latest 64 bit binary from Adobe. With one new Chromium build, it was crashing tabs a lot on flash content. I then noticed that it was (now?) using the mozilla plugin, instead of trying to render flash content natively. (I keep my libflashplayer.so in /home/grogan/.mozilla/plugins and after upgrading it, Chromium was happy with it)

**Dark Angel** · 11-04-2011, 04:27 PM

Gro, I've been trying for a while now to compile the latest (v17.0.927.0) Chromium with no luck. I got past one error with some searching and disabling nacl but I'm getting stuck on something to do with kerebos security ( ./net/http/http_auth_gssapi_posix.h:9:20: fatal error: gssapi.h: No such file or directory )
I reinstalled the package with gassapi.h in it which didn't fix the problem. Searching suggests that using the gnome-keyring would sort this, but I'm not running gnome (I'm using XFCE) and trying to not install gnome (or KDE) packages just willy-nilly.

Not being a master compiler I'm using a slackbuild script. It worked fine for v 11 something, but I had odd things happening here and on other forums, some enough to make the forum unreadable, so I junked it and am trying with newer source.

Here's the slackbuild if it helps:

Code:

 PRGNAM=chromium
VERSION=${VERSION:-17.0.927.0}
BUILD=${BUILD:-1}
TAG=${TAG:-_SBo}


if [ -z "$ARCH" ]; then
  case "$( uname -m )" in
    i?86) ARCH=i486 ;;
    arm*) ARCH=arm ;;
       *) ARCH=$( uname -m ) ;;
  esac
fi

CWD=$(pwd)
TMP=${TMP:-/tmp/SBo}
PKG=$TMP/package-$PRGNAM
OUTPUT=${OUTPUT:-/tmp}

if [ "$ARCH" = "i486" ]; then
  SLKCFLAGS="-O2 -march=i486 -mtune=i686"
  LIBDIRSUFFIX=""
elif [ "$ARCH" = "i686" ]; then
  SLKCFLAGS="-O2 -march=i686 -mtune=i686"
  LIBDIRSUFFIX=""
elif [ "$ARCH" = "x86_64" ]; then
  SLKCFLAGS="-O2 -fPIC"
  LIBDIRSUFFIX="64"
else
  SLKCFLAGS="-O2"
  LIBDIRSUFFIX=""
fi

set -e

rm -rf $PKG
mkdir -p $TMP $PKG $OUTPUT
cd $TMP
rm -rf $PRGNAM-$VERSION
tar xvf $CWD/$PRGNAM-$VERSION.tar.bz2
cd $PRGNAM-$VERSION
chown -R root:root .
find . \
 \( -perm 777 -o -perm 775 -o -perm 711 -o -perm 555 -o -perm 511 \) \
 -exec chmod 755 {} \; -o \
 \( -perm 666 -o -perm 664 -o -perm 600 -o -perm 444 -o -perm 440 -o -perm 400 \) \
 -exec chmod 644 {} \;

# Configure it before compiling
# Change "-Dproprietary_codecs" to "0" if you don't want to use proprietary codecs
#

build/gyp_chromium --depth=. \
 -Dgcc_version=452 \
 -Duse_gconf=0 \
 -Duse_gnome_keyring=0 \
 -Dlinux_link_gnome_keyring=0 \
 -Dremove_webcore_debug_symbols=1 \
 -Dremoting=0 \
 -Dchromeos=0 \
 -Dwerror=0 \
 -Dno_strict_aliasing=1 \
 -Dwerror= \
 -Dlinux_sandbox_path=/usr/lib${LIBDIRSUFFIX}/chromium/chrome_sandbox \
 -Dlinux_strip_binary=1 \
 -Dproprietary_codecs=1 \
 -Duse_system_libjpeg=1 \
 -Duse_system_libxslt=1 \
 -Duse_system_libxml=1 \
 -Duse_system_bzip2=1 \
 -Duse_system_zlib=1 \
 -Duse_system_libpng=0 \
 -Duse_system_yasm=1 \
 -Duse_system_libevent=1 \
 -Ddisable_sse2=${NO_SSE2:-0} \
 -Dffmpeg_branding=Chrome \
 -Duse_system_ffmpeg=0 \
 -Duse_system_ssl=1 \
 -Ddisable_nacl=1


# Compiling...
CFLAGS="$SLKCFLAGS" \
CXXFLAGS="$SLKCFLAGS -fno-ipa-cp" \
make chrome chrome_sandbox BUILDTYPE=Release || exit 1

# Creating the package
cd out/Release/
  mkdir -p $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  mkdir -p $PKG/usr/bin
  install -m 0755 -D chrome $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  install -m 4555 -o root -g root -D chrome_sandbox $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  install -m 0644 -D chrome.pak $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  install -m 0644 -D resources.pak $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  install -m 0755 -D libffmpegsumo.so $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  cp -a locales/ $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  cp -a resources/ $PKG/usr/lib${LIBDIRSUFFIX}/chromium
  find $PKG -name '*.d' -type f -delete

  mkdir -p $PKG/usr/man/man1
  install -m 0644 -D chrome.1 $PKG/usr/man/man1/
  mkdir -p $PKG/usr/share/pixmaps
  cp product_logo_48.png $PKG/usr/share/pixmaps/chromium.png
  mkdir -p $PKG/usr/share/applications
  cp $CWD/chromium.desktop $PKG/usr/share/applications

cd ../../
mkdir -p $PKG/usr/doc/$PRGNAM-$VERSION
cp -a \
  AUTHORS LICENSE \
  $PKG/usr/doc/$PRGNAM-$VERSION/

find $PKG | xargs file | grep -e "executable" -e "shared object" | grep ELF \
  | cut -f 1 -d : | xargs strip --strip-unneeded 2> /dev/null || true

find $PKG/usr/man -type f -exec gzip -9 {} \;
for i in $( find $PKG/usr/man -type l ) ; do ln -s $( readlink $i ).gz $i.gz ; rm $i ; done

cd $PKG
  ln -s /usr/lib${LIBDIRSUFFIX}/chromium/chrome $PKG/usr/bin/chrome
  ln -s /usr/lib${LIBDIRSUFFIX}/chromium/chrome_sandbox $PKG/usr/bin/chrome_sandbox

cat $CWD/$PRGNAM.SlackBuild > $PKG/usr/doc/$PRGNAM-$VERSION/$PRGNAM.SlackBuild

mkdir -p $PKG/install
cat $CWD/slack-desc > $PKG/install/slack-desc
cat $CWD/doinst.sh > $PKG/install/doinst.sh

cd $PKG
/sbin/makepkg -l y -c n $OUTPUT/$PRGNAM-$VERSION-$ARCH-$BUILD$TAG.${PKGTYPE:-tgz}

**Dark Angel** · 11-04-2011, 11:23 PM

Ok, got it to compile, but it has the same issues as the last one. I suppose that means it's probably something it's linked to that's out of date. That makes it sooooo much easier.

**Dark Angel** · 11-05-2011, 12:21 AM

Ok, got FF 7.01 compiled. Not bad and seems to behave better than I'm seeing with Chromium. No fiddling to get it to a 60+ fps result, either.

**Grogan** · 11-05-2011, 12:34 AM

gssapi.h is actually a mozilla header (internal, used to build it). What I did was just copied it to /usr/include from a firefox source tree. It's a mistake that the build doesn't include it.

Here is how I compile Chromium.

First of all, the gyp configuration

See Post #41 for revised procedure. I just run it out of /opt/chromium now with no bin or lib directories. That simplifies things a little

Code:

build/gyp_chromium --depth=. -Dgcc_version=45 -Duse_gconf=0 -Ddisable_nacl=1 -Duse_gnome_keyring=0 -Dlinux_link_gnome_keyring=0 -Dremove_webcore_debug_symbols=1 -Dremoting=0 -Dchromeos=0 -Dwerror=0 -Dno_strict_aliasing=1 -Dwerror= -Dlinux_sandbox_path=/opt/chromium/lib/chrome_sandbox -Dlinux_strip_binary=1 -Dproprietary_codecs=1 -Duse_system_libjpeg=1 -Duse_system_libxslt=0 -Duse_system_libxml=0 -Duse_system_bzip2=1 -Duse_system_zlib=1 -Duse_system_libpng=0 -Duse_system_yasm=1 -Duse_system_libevent=1 -Dffmpeg_branding=Chrome -Duse_system_ffmpeg=0 -Duse_system_ssl=0

Note that I manually install mine to /opt/chromium... use the -Dlinux_sandbox_path to the same as in your SlackBuild, if you're going to use it to build a package.

Note that I am not using system libxml and libxslt. That was the cause of the problems.

I compile it like this:

Code:

time CFLAGS="-march=core2 -fPIC" CXXFLAGS="-march=core2 -fPIC -fno-ipa-cp" make chrome chrome_sandbox BUILDTYPE=Release -j10 V=1

Change -march= to your cpu type. e.g. -march=k8 for an athlon 64 or -march=barcelona for newer ones (see The GCC Manual) Omit the -march= if unsure, or use -mtune= for a safer optimization.

That takes 15 minutes on my box, with all 8 processor threads blazing. Adjust -j to number of CPU threads + 1 or 2 jobs. (e.g. on a dual core processor -j3 or so) Mine is a quad core i7 with hyperthreading and -j10 works best for me.

As I said, I want mine in /opt, I don't really want it installed to the system. I made a script to (re)install mine:

Code:

#! /bin/sh

VERSION=17.0.920.1

mkdir -p /opt/chromium/lib
mkdir -p /opt/chromium/bin
rm -rf /opt/chromium/lib/*
rm -f /usr/share/pixmaps/chromium.png
cd /storage2/shit/build/chromiumbuild/chromium-$VERSION/out/Release
strip -p chrome
strip -p chrome_sandbox
strip -p --strip-unneeded libffmpegsumo.so
cp chrome /opt/chromium/lib
cp chrome_sandbox /opt/chromium/lib
chmod 4555 /opt/chromium/lib/chrome_sandbox
cp chrome.pak /opt/chromium/lib
cp resources.pak /opt/chromium/lib
cp libffmpegsumo.so /opt/chromium/lib
cp product_logo_48.png /usr/share/pixmaps/chromium.png
cp -a locales /opt/chromium/lib
cp -a resources /opt/chromium/lib
chown -R root:root /opt/chromium/lib/locales
chown -R root:root /opt/chromium/lib/resources
cd /opt/chromium/bin
ln -sf ../lib/chrome chrome
ln -sf ../lib/chrome_sandbox chrome_sandbox
cd /usr/local/bin
ln -sf /opt/chromium/bin/chrome chrome
ln -sf /opt/chromium/bin/chrome_sandbox chrome_sandbox

The version string at the top needs to change. (To the same as the build directory)

That reminds me, I haven't done a build in a few days. I guess now is the time.

**Dark Angel** · 11-05-2011, 02:09 AM

Ok, I tweaked the SlackBuild script based on how you do it and I'll see what happens.

**Grogan** · 11-05-2011, 02:32 AM

You probably already know this, but I added (forgot to mention):

Change -march= to your cpu type. e.g. -march=k8 for an athlon 64 or -march=barcelona for newer ones (see The GCC Manual) Omit the -march= if unsure, or use -mtune= for a safer optimization.

**Dark Angel** · 11-05-2011, 02:36 AM

I'm running this on a Core2 Quad Q6600 so what you listed is right for me as is, though the number of threads is different ( -j5 instead of -j10).

I don't worry about adding the "time" par though. As long as it works I don't really care how long the compile takes.

**Grogan** · 11-05-2011, 02:39 AM

Yes, perfect.

By the way, I don't add -O2 to the CFLAGS, because this project uses -O2 for some stuff, -O3 for other stuff, and -Os as well and I see no reason to override their choices. (They ought to know)

**Dark Angel** · 11-05-2011, 02:51 AM

The SlackBuild script I'm using has -o2 for every listed arch as it came. If I have any more issues I'll remove it and try again.

**Dark Angel** · 11-05-2011, 03:18 AM

Ok, it works here now but it's still flakey at the other forum.

Feels quicker than before though.

edit: trying again without the -o2 flag and without system libjpeg to see if I can get it working across all the forums I use.

**Dark Angel** · 11-05-2011, 04:58 AM

Nope. Still does it.

**Grogan** · 11-05-2011, 05:25 AM

Shit, sorry, I could have saved you the trouble. The -O2 flag wouldn't break anything, it would just possibly not be the best thing to do for optimizing the compiled code to override what was intended for a release build. (If anything)

**Dark Angel** · 11-05-2011, 05:38 AM

No matter. It's definitely faster than my firefox build that I optimised for this system. I ran that test page you linked to.
Short results:
User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/535.8 (KHTML, like Gecko) Chrome/17.0.927.0 Safari/535.8
WebGL VENDOR: WebKit
WebGL VERSION: WebGL 1.0 (OpenGL ES 2.0 Chromium)
WebGL RENDERER: WebKit WebGL
Unmasked VENDOR: undefined
Unmasked RENDERER: undefined
WebGL R/G/B/A/Depth/Stencil bits (default config): 8/8/8/8/24/8

-------------------

Test Summary (8521 total tests):
Tests PASSED: 8505
Tests FAILED: 16
Tests TIMED OUT: 1
Overall it worked. My firefox build crashed and burned on the same page.

**Dark Angel** · 11-05-2011, 08:04 PM

Ok, managed to crash one tab in Chromium screwing around on youtube. One tab. Not the whole thing. I like this part. Had a bugger of a time closing the process though ... actually couldn't find it to kill it. Still better than losing the whole thing.

**Grogan** · 11-05-2011, 08:23 PM

Yes, it's worth the extra memory for that alone.

I am not having this browser crash tabs anymore, since upgrading my flash plugin. I have "flashplayer11_rc1_install_lin_64_090611.tar.gz" and I simply unpack it and copy the libflashplayer.so to /home/grogan/.mozilla/plugins. (You don't need any of the other junk that is in the tarball)

Actually, that seems to have gone "final" now. I haven't tried it yet "install_flash_player_11_linux.x86_64.tar.gz" but I'm going to drop it in now.

I don't know for sure whether this link will work or not without a session cookie or something... adobe likes to make downloading difficult (It's much too simple to give anyone an ordinary href URL to click to download something). They have to mystify it with forms and download managers. For Windows boxes that need a flash player upgrade I always just go right to their troubleshooting pages to find direct download links to standalone installers.

http://get.adobe.com/flashplayer/com...tar.gz)_64-bit

**Dark Angel** · 11-06-2011, 02:24 AM

aaaaaand now they're released Firefox 8.0 final. Oh well, another compile to do.

**Dark Angel** · 11-06-2011, 03:46 AM

Done. Latest jre done as well. Still crashes out on the nasty test page though. Seems just slightly quicker on the speed page, getting to 60+ a little quicker than v7.01, but still not as fast as Chromium. DANG this thing honks along!

**Grogan** · 11-06-2011, 03:05 PM

That doesn't impress me that they keep churning out major version number increases, breaking people's extensions etc. It's partly what caused me to investigate an alternative browser and I sold myself on the benefits of Chromium. I now use Firefox only for important things (e.g. server control panels, bank transaction forms etc.) and I probably don't even need to worry about that anymore since I figured out the XML problem.

I am now even allowing Google Chrome to exist on customer computers. I was previously yanking that because it is foistware... gets installed without people really wanting it. In almost all cases it's unused, unless it got made the default and then people were grateful to have it gone. It is actually confusing in its simplicity when people don't think in abstract ways to find the settings.

I guess it's time to upgrade Firefox again though. I haven't done a build since 7.0, and skipped a few versions before that. There just isn't much incentive, when I rarely use it anymore and present versions are fine for those few specific things. I only use two extensions... Video Download Helper (which is very good at staying updated) and Status4Evar, but that one can go. I'm over the status bar now, though I do miss the convenience.

**Dark Angel** · 12-15-2011, 04:47 AM

Well, here we are again I'm screwing with things. Full OS reload after a bunch of annoyances got to me (different thread if I bother starting it) but I'm in Debian 6 and having issues with Chromium. After compiling and installing per the methods in this thread the only way I can get the damn thing to run is to add the --no-sandbox flag. I didn't have to do this before, even on Ubuntu so I figure it's my fault for missing something. It compiles without errors (now) but if I don't use that flag it just won't DO anything but show an error page.
Anyone got any ideas?

Thread: Chromium browser – 636 days old

Thread Tools

Display

Chromium browser

Bookmarks

Bookmarks

Posting Permissions