Wednesday, January 6, 2010

Chrome becomes headache on Linux box




I'll admit it. I had my early misgivings about the Google Chrome browser when it was first coming out. Like, what's Google doing in the software business?

But after using Chrome a few times on Windows systems, it sort of grew on me. It's one fast browser, maybe even faster than Opera. It's less likely to blow up in memory like Firefox. I have a Portable Apps version that I use on borrowed Windows machines, though it doesn't seem to save my bookmarks.

On my Linux box, I'm always experimenting with browsers. I have Firefox, Seamonkey (which is built off the old Mozilla code), Dillo, Lynx, and Opera available to browse the Web, and I use them all. For the past few months Opera has been my go-to, and even that feels poky next to Chrome.

So I put myself on the Chrome mailing list, waiting for the Linux version to come out. A few weeks ago, the first version for MacIntosh and Linux was released, with all the warts.

Now, understand, this version is in beta. Which means it may or may not work. That's what happens when you go for such bleeding-edge software. You're volunteering your services as a crash-test dummy.

So far, only binaries of Chrome seem to be available in Linux; I haven't seen any source-code bundles just yet. I use Vector Linux, which runs with .tlz and .tgz binary packages, but I like to compile my programs from source. Best I was able to do was to download the .rpm binary (which Red Hat and Fedora use), and convert it over to .tgz. Here's what I did, from the root shell:

root:# cd /home/eric
root:# rpm2tgz google-chrome-beta_current_i386.rpm


From there, I was able to install it directly into my system. I was ready to go. I had my seatbelt buckled and my Tony Stewart racing gear on, 'cause this was going to be one fast browser.

A problem, though. I could only go so far; the sandbox (a security tool) wasn't configured right. My command shell told me what to do, so after a few days of deliberation I went ahead with the fix. Here's how, again from a root shell:

root:# su /opt/google/chrome
root:# chown root chrome-sandbox

root:# chmod 4755 chrome-sandbox


Then fired Chrome up from the command shell -- success, for a minute.

As it stands now, I can work Chrome all day, as long as I don't use it to go online. That's when it quits.

Seriously. Here's what I get in the command shell:

eric:$ google-chrome
[7241:7253:703254262:ERROR:/usr/local/google/home/chrome-eng/b/slave/chrome-official-linux/build/src/base/nss_init.cc(89)] Error initializing NSS with a persistent database (sql:/home/eric/.pki/nssdb): NSS error code -8174 Assertion failure: lock != NULL, at ../../../../pr/src/pthreads/ptsynch.c:205 Aborted

In non-technical language, that means it crapped out. Back to the drawing board.

Might need to download the next version, maybe attack one of the nightly builds. And if anyone has any ideas of their own, I'm willing to try them. But despite my early misgivings I see Chrome in my future, and I'm not talking about a chrome dome here.

I'm gonna keep trying, because I'm stubborn and enjoy fooling with software when it's in beta. Face it. I make a good crash test dummy.

###

(You tell me: How about the rest of y'all Linux users? Have you had any better luck than me? Use the comments section below.)

1 comment:

Anonymous said...

It's important to realize that Chrome is (kind of) closed source. It uses the source of Chromium, Google's open source browser project. Chromium is an svn repo with source code and Chrome is the packaged binary version that Google distributes.

You could build Chromium from source, but I've heard that's a really intense process. Gentoo and Arch both offer precompiled packages of Chromium so you might give those a try.

Other than some minor flash issues, I've had good experiences with both Chrome and Chromium on linux.

About YOU