The Blog of Tom Webster
Chronic Ranter, Reviewer, and Developer. I speak only for myself, opinions are my own.
Sometimes, running Debian Unstable is... well... unstable. Most of the time it's perfectly stable, it's just bleeding-edge Debian, perfect for my workstations at home, where I want the latest and greatest the FOSS community has to offer. Rarely, a package will just break or come with some nasty bugs. Two I've gotten bit by recently are network-manager and nvidia-driver. These have been fixed recently, but for a bit, the network-manager bug really bummed me out, I wanted a solution to get back to 'normal' (and I had already cleared my apt-cache).
In both cases, I was able to install the package from stable or
testing and get back to my original functionality. Today I'll walk
you through how to do that.
First, let's make sure we have the ability to use HTTPS with apt:
sudo apt-get install apt-transport-https
I use the https mirrors at kernel.org, so my
/etc/apt/sources.list looks like this:
deb https://mirrors.kernel.org/debian/ sid main contrib non-free
deb-src https://mirrors.kernel.org/debian/ sid main contrib non-free
deb https://mirrors.kernel.org/debian/ testing main contrib non-free
deb-src https://mirrors.kernel.org/debian/ testing main contrib non-free
deb https://mirrors.kernel.org/debian/ stable main contrib non-free
deb-src https://mirrors.kernel.org/debian/ stable main contrib non-free
Next, let's set our default release. This will tell apt where we
prefer our packages to come from. In my case, I want my packages to
come from sid (the codename for Debian unstable).
Add this line to /etc/apt/apt.conf.d/70debconf:
APT::Default-Release "sid";
Note: If you want your default release to be stable or
testing, use that instead of sid in the line above.
What this does is tell apt and dpkg that we prefer for our packages to
come from the sid release channel, that keeps us on the unstable
branch of Debian. Nice and current. To see this in action, run this
command:
sudo apt-cache policy network-manager
network-manager:
Installed: 0.9.10.0-7
Candidate: 1.0.10-2
Version table:
1.0.10-2 990
990 https://mirrors.kernel.org/debian sid/main amd64 Packages
1.0.10-1 500
500 https://mirrors.kernel.org/debian testing/main amd64 Packages
*** 0.9.10.0-7 500
500 https://mirrors.kernel.org/debian stable/main amd64 Packages
100 /var/lib/dpkg/status
See the 990 next to the sid version? That's our pin-priority
value, the highest value wins when apt-get looks for a version to
install. By setting our default release channel, we've set sid's
pin-priority to 990.
For more information on pinning, check out these great resources:
You can now use -t stable to install a package from the stable
repositories:
sudo apt-get install -t stable network-manager
Now, that package will be upgraded next time you run updates (even
unattended-upgrades will do this), so you need to tell dpkg to hold the
package and add it to the "Don't auto-upgrade this one" list:
If you have unattended-upgrades installed and configured, you should
see this section in /etc/apt/apt.conf.d/50unattended-upgrades:
// List of packages to not update (regexp are supported)
Unattended-Upgrade::Package-Blacklist {
// "vim";
// "libc6";
// "libc6-dev";
// "libc6-i686";
};
Add your package to that list:
// List of packages to not update (regexp are supported)
Unattended-Upgrade::Package-Blacklist {
// "vim";
// "libc6";
// "libc6-dev";
// "libc6-i686";
"network-manager";
};
Now unattended-upgrades won't upgrade that package, but you still can
accidentally upgrade it if you use sudo apt-get upgrade or sudo
apt-get dist-upgrade, so here's how to fix that problem:
Run: echo "network-manager hold" | sudo dpkg --set-selections
Note: You don't want to leave packages on hold forever, that would be very very bad. Packages regularly get security updates, and even the bug you suffer from currently will eventually be fixed (watch the Debian bug tracker, it's helpful), so you'll want to remove the hold once a new version of the package has been released. If you keep a package on hold forever, you will compromise the security of your system!
To "unhold" a package, use this command:
echo "network-manager install" | sudo dpkg --set-selections
Now you can sudo apt-get upgrade or sudo apt-get dist-upgrade to
get the latest release of the package.
If you have any questions or comments, hit up the section below.
Sometimes you just gotta fix a date in git commits, and this can happen for a wide variety of reasons. Maybe you were on a system with an unset clock and you want accurate timekeeping, maybe you don't want to admit you were up at 4AM coding. I get it. Here's the fix:
git filter-branch --env-filter \
'if [ $GIT_COMMIT = 119f9ecf58069b265ab22f1f97d2b648faf932e0 ]; then
export GIT_AUTHOR_DATE="Fri Jan 2 21:38:53 2009 -0800"
export GIT_COMMITTER_DATE="Sat May 19 01:01:01 2007 -0700"
fi'
Passphrases are the new hotness. Who cares about upper case, lower case, symbols, and numbers? Not me.
Caveats:
Time to crack is based on a massive cracking array that runs one hundred trillion guesses per second. (Thanks to GRCs password haystacks page for the calculations)
tha2uy2Ieti+ae3chav3Ho{cik6gpahzoon2uCh9phoS'iSeeBaunlock syntax for tailspinunlock syntax for massive director tailspinThe evidence is clear, the four-word passphrase is not only easier to remember, but it's even more secure than the 23-character all-random password. Let's keep this in mind when choosing a new root passphrase.
You should be using a password manager by now. Also, if you're storing the passphrases securely, why not make a 64-character random password for everything? Use a nice passphrase as your master password.
I recently switched to Gnome 3 on my personal laptop. When Gnome 3 first launched, it was an utter train wreck. Completely unusable, poor design choices, and seemingly purposely-built to serve a non-existent straight-linux tablet market. While the tablet influences are still around, the project has moved towards its original roots.
While Gnome 3 still isn't the perfect being that Gnome 2 (now MATE) was, the tweak tool being part of the default install and the extension library are helping make up for the shortcomings. One thing I really wanted to do is add custom startup applications to my session, this is accomplished via the tweak tool for now, since there is no UI for session management (yes, really).
I found
this great article
on how to do this, it involves creating a app.desktop file that can
be seen by the tweak tool. This means that for every command line
startup utility you want run on gnome session start, it needs a
.desktop file. Ridiculous, but it works.
In case the site ever disappears, I've paraphrased the content below:
Create a app.desktop file in ~/.config/autostart, if this
directory doesn't exist, create it.
Here's the basic content of a app.desktop file:
[Desktop Entry]
Type=Application
Exec=/usr/bin/xflux -z 12345
Hidden=false
X-GNOME-Autostart-enabled=true
Name[en_US]=xflux
Name=xflux
Comment[en_US]=Screen color temperature changer
Comment=Screen color temperature changer
Then head over to the Gnome Tweak Tool and go to the Startup
Applications tab. From there, you can select the program for the
app.desktop file you just made. Repeat as necessary.
To show the speed differences in some ruby applications, I wrote a simple ceaser cipher in a few different ways, then timed how long they took to run. I thought the results were interesting enough to post here.
Big thanks to Alan Skorkin for the simple timing code.
The first way creates a string of A-Z, then add the lower case
letters to it, then split each character out, making an array. This
implementation isn't a one-line array creation like the second.
The second way uses a string that's already created with A-Za-z, and
splits them into an array.
The third way uses a hash of a-z, then creates the mappings for
A-Za-z.
The fourth way uses a full hash of A-Za-z.
Let's post some graphs!
I removed the outliers from the four trials and made some graphs. First up, the average for 92 trials:
As you can see, it looks like the full hash implementation is the clear winner here.
Next up is the 92 trials plotted out:
Here's the code if you would like to run your own analysis.
#!/usr/bin/env ruby
# There are lots of ways to do things in ruby. Some are faster than others.
# This ruby script outputs benchmarks in CSV format
# Thanks to Alan Skorkin for the simple timing code
# http://www.skorks.com/2010/03/timing-ruby-code-it-is-easy-with-benchmark/
require "benchmark"
print "HalfArray,FullArray,HalfHash,FullHash\n"
100.times do
# Half array manipulation
def ceaser_cipher(string)
alphabet = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
alphabet += alphabet.downcase
alphabet = alphabet.split(//)
converted_string_array = []
string.split(//).each do |l|
if alphabet.include?(l)
converted_string_array.push(alphabet[(alphabet.find_index(l) - 51).abs].swapcase)
else
converted_string_array.push(l)
end
end
return converted_string_array.join
end
time = Benchmark.realtime do
ceaser_cipher("ABC abc")
end
print (time * 1000)
print ","
# Full array manipulation
def ceaser_cipher(string)
alphabet = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz".split(//)
converted_string_array = []
string.split(//).each do |l|
if alphabet.include?(l)
converted_string_array.push(alphabet[(alphabet.find_index(l) - 51).abs].swapcase)
else
converted_string_array.push(l)
end
end
return converted_string_array.join
end
time = Benchmark.realtime do
ceaser_cipher("ABC abc")
end
print (time * 1000)
print ","
# Half hash dictionary
def ceaser_cipher(string)
initial_dictionary = Hash[
"a" => "z",
"b" => "y",
"c" => "x",
"d" => "w",
"e" => "v",
"f" => "u",
"g" => "t",
"h" => "s",
"i" => "r",
"j" => "q",
"k" => "p",
"l" => "o",
"m" => "n",
"n" => "m",
"o" => "l",
"p" => "k",
"q" => "j",
"r" => "i",
"s" => "h",
"t" => "g",
"u" => "f",
"v" => "e",
"w" => "d",
"x" => "c",
"y" => "b",
"z" => "a"
]
dictionary = {}
initial_dictionary.each do |original, reversed|
dictionary[original] = reversed
dictionary[original.upcase] = reversed.upcase
end
converted_string_array = []
string.split(//).each do |l|
if dictionary.include?(l)
converted_string_array.push(dictionary[l])
else
converted_string_array.push(l)
end
end
return converted_string_array.join
end
time = Benchmark.realtime do
ceaser_cipher("ABC abc")
end
print (time * 1000)
print ","
# Full hash dictionary
def ceaser_cipher(string)
dictionary = Hash[
"A" => "Z",
"B" => "Y",
"C" => "X",
"D" => "W",
"E" => "V",
"F" => "U",
"G" => "T",
"H" => "S",
"I" => "R",
"J" => "Q",
"K" => "P",
"L" => "O",
"M" => "N",
"N" => "M",
"O" => "L",
"P" => "K",
"Q" => "J",
"R" => "I",
"S" => "H",
"T" => "G",
"U" => "F",
"V" => "E",
"W" => "D",
"X" => "C",
"Y" => "B",
"Z" => "A",
"a" => "z",
"b" => "y",
"c" => "x",
"d" => "w",
"e" => "v",
"f" => "u",
"g" => "t",
"h" => "s",
"i" => "r",
"j" => "q",
"k" => "p",
"l" => "o",
"m" => "n",
"n" => "m",
"o" => "l",
"p" => "k",
"q" => "j",
"r" => "i",
"s" => "h",
"t" => "g",
"u" => "f",
"v" => "e",
"w" => "d",
"x" => "c",
"y" => "b",
"z" => "a"
]
converted_string_array = []
string.split(//).each do |l|
if dictionary.include?(l)
converted_string_array.push(dictionary[l])
else
converted_string_array.push(l)
end
end
return converted_string_array.join
end
time = Benchmark.realtime do
ceaser_cipher("ABC abc")
end
print (time * 1000)
print "\n"
end
I'm not entirely familiar with low low level ruby, so I can't offer any grand insights, but I thought it was some decent code and a nice set of graphs.