Browser Stats for December 2007

I’ve been publishing browser stats for FileFormat.Info and December 2007 marks a couple of milestones:

• Firefox (all flavors) passed Internet Explorer (all versions).
• IE7 passed IE6.

I suppose FileFormat.Info is a fairly technical site, so this may not be representative of the internet population as a whole.

I am not doing anything browser-specific so it doesn’t really affect me directly, but it is good to see real competition.

Using xsane with a Canon CanoScan LiDE 60 on Debian Etch

I got an inexpensive flatbed USB scanner and was disappointed with the bundled (Windows-only) software, so I tried it on Linux. Debian Etch installs xsane by default, but it would not recognize the scanner. I tried a whole bunch of things but nothing help, until I accidentally started xsane from a root terminal, and it “just worked”, despite the warning from xsane about running as root.

Resuming an HTTP download with Curl

I had to download an ISO image to test some new firewall software and it got interrupted halfway through. Fortunately, curl has a way to continue from where you left out with the “-C -” option. A full command line looks like:

curl -C – –compressed –output disk.iso http://example/download/disk.iso

How to make a word list from an aspell dictionary

I am working on something that needs a list of words, without regard to american vs. british or accents or anything: just has to be as many words as possible.  There are a whole bunch of aspell dictionaries available.  First expand the files:

preunzip *.wl

Then merge into a single list, eliminating duplicates:

sort –unique –ignore-case *.wl >list.txt

In additon, I want everything to be UTF-8:

iconv -f ISO8859-1 -t UTF-8 list.txt >ulist.txt

Pretty simple.  The merged english word list has 137,883 words.

Stop GEdit opening FTP files as read-only

I use Debian’s “Places -> connect to server…” dialog to setup connections to a couple FTP servers. Copying and such like work fine, but when I try to use GEdit, it always opens the file read-only. I can “Save As” to the exact file, so it wasn’t a permission problem. The trick is in some totally obscure config file. Here’s the (in GUI) steps to fix it:

1. Run gconf-editor
2. Open the tree to /apps/gedit-2/preferences/editor/save
3. Edit writable_vfs_schemes
4. Add ftp to the list

Voila. Gedit saving FTP. Via the kaylus on the Ubuntu Forums, but it is the same in Debian Etch.

VMWare Server on a headless Debian Etch

I recently had to install VMWare Server on a server that didn’t have the GUI bits installed.  This is the list of additional packages that need to be installed for VMWare to install:

apt-get install psmisc libxrender1 libxt6

Then it will install, and the VMs can have GUI even though the host doesn’t!

Avoid hogging bandwidth with curl’s –limit-rate

At work we have a slower (1.1 Mbps) internet connection and I don’t want to hog it when I need to download a big file. Fortunately, curl has a –limit-rate option. For example (should be one line):

curl –verbose –output bigfile.tar.gz –limit-rate 3000 http://download.example.com/bigfile.tar.gz

Making PDFs on Debian Etch

This are my notes on setting up a PDF printer on Debian Etch for the local machine (i.e. not as shared printer).

Install cups-pdf:

sudo apt-get install cups-pdf

Add a printer. I find that the browser-based configuration works better than the GUI versions (Foomatic-GUI and Gnome CUPS Manager):

• browse to http://localhost:631/
• Add a printer
• Choose the type Generic postscript color printer rev4
• Choose the location cups-pdf:/

The PDF files are put in a PDF subdirectory of your home directory (~/PDF). However, if you print while logged in as root (for example, the test page from the web interface), the PDFs usually end up in /root/PDF but once I found it in /var/spool/cups-pdf/ANONYMOUS.

One other thing to note is that it uses the title of the print job as the file name. The title is set by the program that’s printing, but if it is there is an existing file with the same name, it will overwrite it silently. Generally this isn’t an issue, but Firefox/Iceweasel uses the page <title> as the print job name, so if pages have generic/bad titles, you have to be careful.

Making images of floppy disks and CD-ROMs on Linux

I have several books that came with floppy disks or CD-ROMs. I’m worried about whether they’ll last and wanted to make copies of them. It is trivially easy on Linux:

dd if=/dev/scd0 of=/iso/book.iso

I got a rawread script that supposed solves some compatibility problems. I found that dd worked fine on two disks that the script had problems with. The script is called with

rawread.sh /dev/scd0 >book.iso

To mount a image (mkdir /media/iso directory first):

sudo mount -o loop book.iso /media/iso

For floppies, use /dev/usb0 (I have a usb floppy drive) and .img as the extension.

Swapping columns in a text file

An awk snippet to swap columns in a text file:

awk -F'[ ]' '{ print $2, ": ", $1 }' original.txt >swapped.txt

The '[ ]' is the separator in the original file (a regular expression). The ": " is the separator that is put in the result file.

You can also be done with cut and paste command line tools.