Jim's Depository

this code is not yet written

Certain corrupt JPEG files will explode Quicktime Player if it encounters them in an “Import Image Sequence…” operation. This bug has been present for many years and shows no sign of going away.

If you are archiving images from cheap webcams you will encounter these.

I am not aware of anything included with Mac OS X Leopard that validates a JPEG file, so I built jpeginfo for Leopard. 

With that, you can do things like… for v in *.jpg ;do jpeginfo -c \$v || rm \$v ;done … to delete all the corrupted JPEGs.

You can find the sources at http://www.iki.fi/tjko/projects.html You will also need to build libjpeg, from ftp://ftp.uu.net/graphics/jpeg/jpegsrc.v6b.tar.gz

Build libjpeg first (./configure ; make ), then build jpeginfo (./configure –with-libjpeg=THE-RIGHT_DIRECTORY ; make )

But you don’t need to do that… I’ve attached the copy I built.


jpeginfo 103388 bytes
I might mention that there are still some jpegs that will crash the quicktime encoder that are valid according to jpeginfo, but it gets most of them.
I found a much simpler solution, there is a tool 'Corrupt JPEG Checker' for Mac OSX.
It can be found here: http://www.failedinsider.com/corruptjpegcheckermacosx/ or in the mac app store.

Saved me a lot of time.

unfortunately JPEGInfo will not catch all.

the tool above catches the rest

Safari is odd with contentEditable divs. It doesn’t assume you will have a pre-formatted div as a container so it puts each line into its own div… sometimes marked with code class, sometimes not.

That makes a mess but is tolerable, until my HTML sanitizer burps out the clean version with newlines between the divs and causes accidental double spacing.

I could rewrite the ->saveHTML() method of the PHP DOMDocument to not put newlines in between divs, but that would give awful looking HTML.

For now I added a function to the sanitizer to remove extra divs from inside code formated divs. It tries to be smart about inserting newline characters, but it may not be smart enough.

Someone should revisit the whole contentEditable thing and specify precisely what is to meant by all of the operations.

If you run your own bind server for your domain you can easily support dynamic updates from your machines that have transient IP addresses. bind is capable of many things, but I’ll just show the bits you need…

First, use bind 9.3 or better.

On the server, go edit the file that contains your zone… perhaps it looks like this… zone “studt.net” { type master;         file “/etc/bind/studt.net”; };

… you are going to need to generate a key for each host that has a transient address, add it to this file, and then tell the zone that machines are allowed to update their own addresses.

dnssec-keygen -a HMAC-MD5 -b 512 -n HOST jimshouse.studt.net

… will generate two files with long names. Keep them, you will need them on the client machines. But look inside the K*.private file and copy out the value of the “Key: “ line. You are about to paste it into your zone file.

Add a key for each host that will update its name, and an update-policy to the zone, you may end up looking like this…

key jimshouse.studt.net {         algorithm HMAC-MD5;         secret “eQGH–lots-of-more-key-i-left-out==”; };
zone “studt.net” {         type master;         notify yes;         forwarders { };         file “/etc/bind/studt.net”;         update-policy {                 grant * self * A TXT;         }; };

Restart your bind server and your server is ready to go. (Yes, you could reload but there was a bug in 1999 or so and I have never gotten over it.)

Possible Bug: Your bind process has to be able to write “.jnl” files. Debian etch is configured to put them in /etc/bind but the bind user can’t write there. I chmodded /etc/bind to 775 to deal with that. You’ll know you have this problem when your client update fails with a SERVFAIL and you tail your syslog on the server and read the error messages.

Dangerous Note: Before you edit the zone file you have to first stop the dynamic updates so the .jnl file gets merged with the zone file… rndc freeze studt.net edit the studt.net zone file rndc unfreeze studt.net
Now for the client side. You could set dhclient.conf to do this automatically, but for primitive cave programmers like me you can just execute a command. First… copy your K*.private key file for the client to the client machine, you’re going to need it. Second… use the nsupdate command to set the name’s value. I do it in a script sort of like this…

#!/bin/sh TTL=600
SERVER=ns.studt.net. ZONE=studt.net HOSTNAME=jimshouse.studt.net. KEYFILE=/path/to/where/you/keep/**Kjimshouse.studt.net.+157+26806.private IP=
nsupdate -v -k \$KEYFILE > /dev/null << EOF server \$SERVER zone \$ZONE update delete \$HOSTNAME A update add \$HOSTNAME \$TTL A \$IP send EOF See that TTL? That says 10 minutes, so a computer on the internet might keep using your old address for up to 10 minutes after your address changes. You can adjust that number for your situation. Clients won’t like you if you get too short.

I’d just like to mention that I would feel better if Tyrannosaurus Rex had lived in Europe. 1,000 miles and 65 million years is too close. I want there to be an ocean between them and me as well.

The first glorious ball and stick iMacs can no longer run current Mac software, you can force it, but it is painful and has a video defect coming back from sleep. So, what do to with them…

They are too pretty to throw out, so turn them into picture frames.

  1. Install Debian Lenny. (burn CD, boot with ‘C’, do the normal stuff)
  2. Try to find a package to just fill the flipping console with a jpeg. Fail.
  3. Write a little program using the fbdev and libjpeg.
  4. Agonize of over “doing the right thing” with image scaling.
  5. Discover freeimage. Pitch previous code, let library do the right thing.
  6. Fail miserably on the iMac. Clearly I am missing something, but I can’t get the console into true color mode. I can make it be 32 bits per pixel in the frame buffer, but it still does some sort of pseudocolor.

It’s quick! It’s not dirty!

It is HUGE! freeimage is a 6M install for my 18K program. That’s ok on this recycled machine with 512M Ram and 40G disk, but I’ll need to code my own Lanczos resampler if I want to use this on small machines.

I generally forget some nice packages when I toss up a new debian machine and then spend too much time trying to remember which ones they are. Now I keep them listed here. Maybe you will like them too.

  • dpkg-dev-el  (for debian/changelog files, stupid picky date syntax)
  • dfu-programmer (to reflash Atmel microcontrollers)
  • iproute (aways forget it)
  • psmisc (killall)
  • sysstat (iostat)
  • strace (sometimes you have to know)
  • tcpdump
  • mtr-tiny (mtr, but without all the X dependencies)
  • ifstat
  • iftop
  • iotop
  • htop,atop (not comitted to these yet, but they seem useful)
  • attr (see and set file’s extended attributes)
  • emacs-goodies-el (markdown mode)

That attachment is an etch build of dfu-programmer, I can’t test it, but it probably works on Etch.



can you tell me how to install dfu-programmer on debian etch?
i can not find the package with apt.

greetings remo
I just built dfu-programmer for etch with these steps, and attached it to the parent article (I guess I should implement attachments for comments.)
  1. Go to the Debian packages page for the Lenny package.
  2. Download the original source and debian diffs.
  3. Unpack the original sources
    tar -zxvf dfu-programmer_0.4.3.orig.tar.gz
  4. Apply debian patches... 
    zcat dfu-programmer_0.4.3-1.diff.gz | patch -p0
  5. Hop in
    cd dfu-programmer-0.4.3
  6. Build packages
    fakeroot ./debian/rules binary-arch
  7. Fail... fix x flags on rules
    chmod +x debian/rules
  8. Fail... install libusb-dev
  9. Success! I now have a dfu-programmer_0.4.3-1_i386.deb in the parent directory.
  10. I can't test it, my only Etch machines are remote servers, but it installs ok and I expect it works.
I probably should have modified the version number to mark it as an Etch build, but this works for private use.