I Make Things - Computing

Run Script at Mac OS X Login

nospam@example.com (Latcarf) — Wed, 06 May 2009 03:28:12 -0400

I know alot of people want to know how to do this. The solution is usually creating a shell script and naming it with the '.command' extension so it'll launch the Terminal.

The cleaner way is through a simple launchd plist placed in your ~/Library/LaunchAgents folder.

Here's my example:

$ cat ~/Library/LaunchAgents/com.latcarf.loginscript.plist

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>com.latcarf.loginscript</string>
<key>ProgramArguments</key>
<array>
<string>/bin/bash</string>
<string>/Users/latcarf/Documents/script/loginscript.bash</string>
</array>
<key>RunAtLoad</key>
<true/>
</dict>
</plist>

# Start it up
$ launchctl load ~/Library/LaunchAgents/com.latcarf.loginscript.plist
# Or just reboot or logout and back in to kick it off.
# Only runs once at login.

It's not MAC!

nospam@example.com (Latcarf) — Mon, 16 Mar 2009 01:43:51 -0400

Just a little pet peeve I need to get off my chest.

I know all you PC folks are coming over to the Mac in droves and that's a great thing. It's great for you and for me. You'll have a much better computing experience (and life in general) on Mac and I'll still have a job making software for you. So win-win all around!

One little behavior we need to correct though, is how you write about your new computer. In the Windows world it's normal to write PC, or DELL, or IBM in all capital letters. Heck, in DOS (all caps there too) it's even proper to write FOLDER NAMES in capital letters as well, isn't it?

In the Mac world however, it's wrong to write Mac in all caps (and generally wrong to write folder names in all caps - it just looks like you're yelling). Yeah, I know it's a hard habit to break; writing Mac vs. MAC. I mean, you're gonna have to hold down the shift key a little less, and that's tough, but you should be getting used to doing things with less effort now that you have a Mac, right? Besides, you don't need to yell about your Mac. We all know how much better they are (we've been using them for years) and Mac users generally appreciate a little more subtly (you'll understand that as you get used to using your new Mac).

Anyway, Mac is the first three letters in the word Macintosh, which is a kind of an apple. Apple makes computers (and other hardware) and a Mac is a kind of Apple computer. See the correlation here, kind of witty huh? Anyway, you never see the word Macintosh with all caps for the first three letters: MACintosh. That just looks weird right? Well, using MAC to talk about Mac computers looks weird too. So, if you could just please not do that, it would be great.

Oh, before you go there is one instance where you'll want to use the all capitals version of the word MAC. That's when you need to talk about the "quasi-unique" identifier that's assigned to your network adapter. It's known as a Media Access Control address. All computers that can connect to a network have these and they're certainly not limited to Mac. Hey, I guess you also might want to use the all capitals version MAC when you're talking about MAC cosmetics as well, but that has nothing to do with computers.

Ok, so just to make sure we're on the same page... Use Mac and not MAC and we'll all get along great. Thanks!

Fink on Leopard With Blocked CVS Port

nospam@example.com (Latcarf) — Tue, 26 Feb 2008 20:05:47 -0400

Recently I wanted to use Fink on my Leopard machine at work. This sounds simple but there are a few complicating factors. First, there are no Fink binaries for Leopard as of this writing. Ok, so that's not a big deal, I can compile it easy enough, Fink made an easy to use bootstrap script for that purpose. Secondly, as far as I know, as of this writing, we can't use rsync mode during Fink's self update on Leopard. So, the solution is to use fink selfupdate-cvs. This leads to the third problem which comes from the fact that my company is strict about blocking ports. Unfortunately the cvs port 2401 is blocked. Fortunately however, SSH (port 22) is not. So my solution is to temporarily port-forward cvs through ssh to my home server so I can run fink selfupdate-cvs to pull down all the great package descriptions I need for my fink install.

Here's are my notes on what I did:

# I needed to find where the cvs urls were set so I grepped in /sw
cd /sw
grep -R fink.cvs.sourceforge.net *

# That eventually turned up the following potential places to make changes:
lib/fink/URL/anonymous-cvs:fink.cvs.sourceforge.net:/cvsroot/fink
lib/fink/URL/cvs-repository:fink.cvs.sourceforge.net
lib/fink/URL/developer-cvs:fink.cvs.sourceforge.net:/cvsroot/fink
lib/perl5/Fink/SelfUpdate/CVS.pm: my $cvsrepository = "fink.cvs.sourceforge.net:/cvsroot/fink";

# Since I only needed to get package descriptions I figured /sw/lib/fink/URL/anonymous-cvs
# was the file I wanted change to handle my port-forward solution.
cd /sw/lib/fink/URL/
sudo cp anonymous-cvs anonymous-cvs.orig
nano -wz anonymous-cvs

# Here I changed: fink.cvs.sourceforge.net:/cvsroot/fink to: localhost:/cvsroot/fink

# Then it was a simple matter to create an ssh tunnel to an external server where I have
# the ability to login and its port 2401 is not blocked.
ssh login@domian.example -L 2401:fink.cvs.sourceforge.net:2401

# Finally, I ran the fink update command:
fink selfupdate-cvs

# Hooray. Now everything works as expected and I probaly don't need to use it again unless I need
# to update package descriptions sometime in the future. So, I changed things back:
cd /sw/lib/fink/URL/
sudo mv anonymous-cvs anonymous-cvs.custom
sudo mv anonymous-cvs.orig anonymous-cvs

# Probably I should create a shell alias like: finkselfupdatecvs to handle it all when I
# I find I want to update package descriptions again. But I'm too lazy and besides, that's
# why I've put it in my blog.

Fix for Tor on Leopard

nospam@example.com (Latcarf) — Mon, 03 Dec 2007 04:07:44 -0400

After installing Mac OS X 10.5 (Leopard) fresh (I won't go into details ), I needed to reinstall Tor. I found the Tor expert installer didn't quite do what it was supposed to as far as creating the _tor user the daemon runs as (the user ends up a little mangled - check it out after install with dscl . -read /Users/tor) Since Leopard doesn't use NetInfo (niutil) anymore, I had to manually create the user with DirectoryServices (dscl). Here's the quick-and-dirty on what I did...

# Create Tor Group and User:
sudo dscl . -create /Groups/_tor
sudo dscl . -create /Groups/_tor PrimaryGroupID 100
sudo dscl . -create /Groups/_tor RealName "Tor Group"
sudo dscl . -create /Groups/_tor Password '*'

sudo dscl . -create /Users/_tor
sudo dscl . -create /Users/_tor UserShell /usr/bin/false
sudo dscl . -create /Users/_tor RealName "Tor Server"
sudo dscl . -create /Users/_tor UniqueID 100
sudo dscl . -create /Users/_tor PrimaryGroupID 100
sudo dscl . -create /Users/_tor NFSHomeDirectory /Library/Tor/var/lib/tor
sudo dscl . -create /Users/_tor Password '*'
sudo dscl . -append /Groups/_tor GroupMembership _tor

After that, there was still a little more to do to get Tor to run. I needed to create a place for the logs and fix the StartupItems script. Here's what I did for that...

# Create the logs directory
cd /var/log
sudo mkdir tor
sudo touch tor/tor.log
sudo chown -R _tor:_tor tor

# Fix perms on the "home" directory
cd /Library/Tor/var/lib
sudo chown -R _tor:_tor tor

# In /Library/StartupItems/Tor/Tor change TORGROUP from TORGROUP=daemon to:
TORGROUP=_tor

# Start the Tor server
sudo /Library/StartupItems/Tor/Tor start

Cool. That fixed it. Still not sure whether I want logging in /var/log/tor or /Library/Tor/var/log/tor though... Also, I guess I probably should submit all this to the Tor folks too. I'm so damn lazy.

Software I Use

nospam@example.com (Latcarf) — Thu, 13 Sep 2007 15:58:01 -0400

I often get people asking me what Mac software I use alot and if there's anything I can recommend to them. I recently put together a list of stuff I use at work (this doesn't include the *NIX stuff) for a colleague and figured others might benefit from it.

Adium - AIM, MSN, Yahoo, .Mac, Jabber, Gadu-Gadu, ICQ, etc... IM client. (open source)
Cyberduck - FTP client (open source)
ClamXav - Virus Checker (not like you'll need it) (open source)
Firefox - Web browser (open source)
iTerm - Terminal Emulator (open source)
Microsoft Remote Desktop Client - RDC to connect to Windows on Mac (free)
VLC - Media Player (open source)
Yojimbo - Password/etc DB (commercial)

TkCVS in Mac OS X X11

nospam@example.com (Latcarf) — Wed, 27 Jun 2007 20:44:38 -0400

TkCVS (TkSVN) is one of the best tools I've found for visualizing CVS and Subversion branches. Since it is written in Tcl/Tk you can use it on Mac, Linux, and that other OS.

Here's how to install it on Mac for use in the X11 environment and how to integrate it with BBEdit. This assumes you have the X11 environment and BBEdit command line tools (bbedit and bbdiff) installed. Also, you should have a /usr/local directory and it has been added to your $PATH. You'll have to find out how to do this on your own if you don't already know how.

Download the Unix distro (the latest was 8.0.4 as of this writing) to your Desktop.

Unpack the distro.

tar -xvzf tkcvs_8_0_4.tar.gz
Install into /usr/local

cd tkcvs_8_0_4
sudo ./doinstall.tcl -nox /usr/local
Launch X11 and launch TkCVS from the xterm. Or, better yet, customize the X11 "Applications" menu and add a command for tkcvs (I set mine to command-t).

tkcvs
Edit the cvsdiff script in the contrib directory.

bbedit contrib/cvsdiff

Find the line: VDIFF="gvim -d -f" and replace it with:

VDIFF="bbdiff --wait --resume"
Copy the cvsdiff script in the contrib directory to /usr/local/bin.

sudo cp contrib/cvsdiff /usr/local/bin/cvsdiff
Edit the .tkcvs config file to use the diff wrapper script and bbedit for comparing and editing.

bbedit ~/.tkcvs

Add the following lines to the config file.

set cvscfg(editor) "bbedit --wait --resume"
set cvscfg(tkdiff) "/usr/local/bin/cvsdiff"
Launch TkCVS again via X11 xterm and you're good to go!

tkcvs

Oh yeah. Don't forget that to cut/copy/paste (etc...) you need to use the control key and not the command key. You're in X-Windows after all.

Vana'diel Time

nospam@example.com (Latcarf) — Mon, 18 Jun 2007 19:50:31 -0400

I recently started working on a Vana'diel (Final Fantasy XI) clock written in Cocoa. My reasoning for doing this (when there really isn't any need for another Vana'diel clock) is two-fold:
1. I want to learn Cocoa and a good way to learn something is to associate it with something you already enjoy (ie. Final Fantasy).
2. The only Mac OS X Vana'diel clock out there is old, lacks some info I'd like it to have, and doesn't really have the nicest interface (text only - no eye candy).

The biggest hump for me so far was figuring out how to calculate the time. I looked at some javascript clocks people had written but I didn't like the fact that they all seem based on some arbitrary date. I wanted something that could go all the way back to the beginning of Vana'diel time. The Vana'diel epoch (the Vana'diel big bang, if you will). I decided, I wanted to write the algorithim myself, differently from all the ways I've seen it done. Fortunately, I found a great resource explaining some of the critical information needed to understand how Vana'diel time works. Thanks Miwarre!

I wanted to rapidly prototype a simple clock to test my algorithm and I figured I could do it easily and quickly enough using Applescript. I've posted the script here for anyone else trying to work with Vana'diel time.

When I re-wrote this for Cocoa, I didn't use the "seconds since epoch" constant I used here because I found a cleaner way with NSCalendarDate. The jist of it is that I can create an NSCalendarDate object using the "earth" date that corresponds to the Vana'diel epoch date and the effect is the same.

I have to give a big special thanks to my girlfriend who checked my work (by hand!) and found that I'd originally used the wrong value in the unix_seconds_since_epoch_at_vana_epoch (it was missing the extra 86400 seconds for the 1968 leap year). This has since been corrected in this script.

The Applescript simply returns the current Vana'diel date.

set unix_seconds_since_epoch to do shell script "date \"+%s\""
set vana_rate_mod to 25 -- Vana time runs 25 times faster than earth time.
set vana_seconds_per_day to 86400
set vana_days_per_week to 8
set vana_days_per_month to 30
--This value is derived from the amount of seconds that elapsed between the
--Vana'diel epoch (0001/01/01 00:00) which is 1967/02/09 15:00 and the
--UNIX epoch which is 1970/01/01 00:00. This includes a leap year (1968/02/29)
--which adds an extra 86400 seconds. Since all calculations are based off seconds
--since the UNIX epoch, we need to add this value to any UNIX seconds we use.
set unix_seconds_since_epoch_at_vana_epoch to 91270800

-- Find out the Vana'diel minutes since the Vana'diel epoch.
set vana_seconds_since_epoch to (unix_seconds_since_epoch + ¬
unix_seconds_since_epoch_at_vana_epoch) * vana_rate_mod

set vana_days_since_epoch to ¬
(vana_seconds_since_epoch / vana_seconds_per_day)
set vana_weeks_since_epoch to ¬
(vana_days_since_epoch / vana_days_per_week)
set vana_months_since_epoch to ¬
(vana_days_since_epoch / vana_days_per_month)

set v_year to round (vana_days_since_epoch / 360) rounding up
set v_month to round ((vana_months_since_epoch) mod 12) rounding up
set v_day to round (vana_days_since_epoch) mod 30 rounding up

set v_hour to round ((vana_seconds_since_epoch mod ¬
vana_seconds_per_day) / 3600) rounding down
set v_minute to round (((vana_seconds_since_epoch mod ¬
vana_seconds_per_day) mod 3600) / 60) rounding down
if v_minute is less than 10 then
set v_minute to "0" & v_minute
end if
set v_second to round (((vana_seconds_since_epoch mod ¬
vana_seconds_per_day) mod 3600) mod 60) rounding down
if v_second is less than 10 then
set v_second to "0" & v_second
end if

set v_wday to round (vana_days_since_epoch mod vana_days_per_week) rounding up

set week_days to {"Firesday", "Earthsday", "Watersday", "Windsday", ¬
"Iceday", "Lightningsday", "Lightsday", "Darksday"}

return (v_year & "/" & v_month & "/" & v_day & " " & ¬
v_hour & ":" & v_minute & ":" & v_second & " " & ¬
item (v_wday) of week_days) as string

Download all Space.com Wallpapers

nospam@example.com (Latcarf) — Mon, 09 Apr 2007 01:01:17 -0400

This is a really quick and dirty script I banged out in about a half-hour to download all the wallpapers at Space.com. I like reading the articles at Space.com from time to time and I found the wallpapers there are pretty cool. I wanted to grab them all and use them for my Mac OS X screensaver. Of course, downloading them individually (click, click, click, right-click, save as, etc... - ad infinitum), is a pain in the ass. So, the Space.com Wallpaper Download (sdcwd) script was born!

The script recursively downloads all the images of the specified size, creating folders for each collection. Just start it and let it run!

I used the following software and versions, you'll need '-o' support in grep for sure and have curl and wget installed. Of course, if the urls change at Space.com, this script will be broken. Your mileage may vary.

Software Versions:
curl - curl 7.13.1
grep - grep (GNU grep) 2.5.1
wget - GNU Wget 1.8.2

#!/bin/bash
# Space.com Wallpaper Download (sdcwd)

image_size="1280" # Set this to 640, 800, 1024, or 1280 depending on the image size you want
collections=(`curl -s http://www.space.com/php/multimedia/downloads/wallpapers/collection.php?collection=adrian_lark | \
grep -o 'option value="collection.php?collection=$.*$"' | \
sed 's,option value="collection.php?collection=$.*$",\1,'`)
collection_url="http://www.space.com/php/multimedia/downloads/wallpapers/collection.php?collection="
image_url="http://a52.g.akamaitech.net/f/52/827/1d/www.space.com/entertainment/downloads/spaceart/images/"

logFile="sdcwd.log"
topDir="Space.com_Wallpapers"

getImages()
{
this_collection="$1"
collurl="${collection_url}${this_collection}"

if [ ! -d "$this_collection" ]; then
mkdir "$this_collection"
cd "$this_collection"

# Log
echo "----------------------------" >> "../$logFile"
echo "$this_collection" >> "../$logFile"
echo "----------------------------" >> "../$logFile"

curl -s "$collurl" | grep "/php/multimedia/imagedisplay/wallpaper_display.php?pic=.*_$image_size\.jpg" | \
grep -o "pic=$.*.jpg$" | \
sed "s,pic=,$image_url," | \
tee -a "../$logFile" | \
xargs wget -q

# Log
echo "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~" >> "../$logFile"
echo >> "../$logFile"

cd ../
# If we didn't download any pics for this directory (maybe the requested size wasn't available)
# remove the empty directory.
find "$this_collection" -type d -empty | xargs rmdir
fi
}

old_dir="`pwd`"
mkdir "$topDir"
cd "$topDir"
for this_collection in ${collections[@]}
do
getImages "$this_collection"
done
cd "$old_dir"

Recursively Convert Line-Endings

nospam@example.com (Latcarf) — Wed, 07 Mar 2007 22:15:32 -0400

$ find . -type f -exec perl -pi -e 's/\r\n?/\n/g' {} \; ¬

Found on Madness & Macness

OvoLab Phlink

nospam@example.com (Latcarf) — Fri, 26 Jan 2007 21:49:25 -0400

I purchased and finally received an OvoLab Phlink. I've connected it to my Mac Mini (my home automation server) and haven't gotten much further than running the software and calling it a few times. Eventually, I plan on setting up full control of my home X-10 system but of course, like anything fun, it takes some time to do it.

So far I've noticed some issues with hanging up and not leaving a message. Phlink seems to leave the telephone connected and never really hangs it up. I'm hoping it will be a transient problem - the kind that only happens when you're playing around with something that isn't set up properly yet. I noticed a few reports of the same behavior on the Phlink forum but never really saw a definitive solution. Further experimentation will certainly be called for.

Hopefully I'll get everything set up to the point that I have total control over my house from a few button presses on my cell phone! Isn't technology great?

Upgrade Hard Drive on Debian Linux (Sarge)

nospam@example.com (Latcarf) — Sun, 10 Sep 2006 23:23:43 -0400

This weekend I decided to upgrade my Linux server (the same server running this weblog). The machine is a decommissioned corporate Compaq DeskPro EN - 1GHz Pentium III with a 20GB hard drive. When I first installed Debian 3.1 (Sarge) I installed it on the original hard drive. I ran it that way for over two years and finally decided to replace the hard drive with a 300GB one. After doing some Googling on the subject, it seemed like this might not be too hard. Then, this is Linux after all and even the simplest tasks can balloon in complexity depending on the hardware you're working with. As I worked through the task it seemed a little tough but, in retrospect, it really wasn't that problematic. Here's what I did...
Continue reading "Upgrade Hard Drive on Debian Linux (Sarge)"

Clamcorder Released!

nospam@example.com (Latcarf) — Sun, 04 Jun 2006 23:33:48 -0400

There's a mini-game in Final Fantasy XI called "clamming." You dig up items in the sand and drop them into a bucket. Each item has a specific weight and the bucket can only hold so much before it breaks. If you can fill the bucket within five ponzes (the weight unit of Vana'diel - the world in Final Fantasy XI), you can transfer your items in to a larger bucket and continue digging stuff up. If your bucket breaks, you lose anything you've dug up so far. At anytime you can stop clamming and save the items.

The mini-game is appealing because it's got a mixture or skill and chance some items you might find can be worth alot of gil (the monetary unit in Vana'diel). I started clamming for fun and then I got hooked on it and decided I could start making some profit. In the course of clamming, I created a tool that helps me track the weight in my bucket. As things go, it ended up being able to do much more than that.

Today, I finally released it to the public as a free download! It's called Clamcorder (a play on "clamming" and "recorder").
Continue reading "Clamcorder Released!"

Convert Date to UNIX Timestamp

nospam@example.com (Latcarf) — Tue, 09 May 2006 13:39:28 -0400

I looked all over trying to find an easy way to convert a date to a UNIX timestamp in a shell or perl script. I was disappointed that I didn't find anything that was easy, sure and didn't require extra software. Most of my UNIX scripts have to operate on Mac OS X so I don't concern myself too much about portability and generally try to use what's already available on the system.

The easiest way I found to convert a date string to a UNIX timestamp is with PHP's strtotime function. Fortunately, PHP is included on Mac OS X machine at no extra charge! This one-liner handles the need pretty well.

$ my_stamp=`php -r \ ¬
'fwrite(STDOUT,strtotime("2006-05-08 15:23:52"));'` ¬
$ echo $my_stamp ¬
1147127032

The constants: STDIN, STDOUT, and STDERR are predefined to make life easy!

For more info, check out: "Using PHP from the command line."

An Easy SQL Search and Replace

nospam@example.com (Latcarf) — Thu, 04 May 2006 13:25:21 -0400

After making some path changes this blog here, I needed to update some of the URLs in previous entries (mainly image URLs). I thought it was going to be tough but it turns out there's an easy way to do it with a SQL one-liner.

UPDATE myTable SET myColumn = REPLACE(myColumn, 'searchText', 'replaceText');

I love simple solutions!

Recursively Delete Subversion Properties

nospam@example.com (Latcarf) — Fri, 14 Apr 2006 19:08:47 -0400

My conversion from cvs to svn caused a bunch of files to end up with svn:keywords set for Author, Date, Revision, and Id. I recursively deleted the svn:keywords property from all the files with the following command. This could work with other svn property commands as well.

$ cd ~/sandbox/project/ ¬
$ find . ! -path "*svn*" \
> -exec svn propdel "svn:keywords" "{}" \; ¬