Most of the info I found on the web didn’t help, but I eventually solved it myself by using hook post-revprop-change instead of post-commit with the code shown below. This code only activates when svnsync is completely finished and has unlocked the destination repository, then it syncs the latest revision with Trac and then runs the Trac post-commit hook to close tickets, etc. Svnsync only seems to update commit messages and users after committing code, so Trac has the wrong information in post-commit, so you have to use post-revprop-change instead.

REPOS="$1"
REV="$2"
USER="$3"
PROPNAME="$4"
ACTION="$5"

# Export bash variables required by trac.fcgi
export HOME="/home/me"
# Root directory for the Trac project
export TRAC_PROJECT_DIR="/home/me/trac_projects"
export TRAC_ENV="/home/me/trac_projects/myproj"

. ${HOME}/.bash_profile

if [ "$ACTION" = "D" -a "$PROPNAME" = "svn:sync-lock" ]; then
    YOUNGEST_REV=`svnlook youngest /home/me/svn/myproj`
    trac-admin $TRAC_ENV resync $YOUNGEST_REV
    ${RUN}/bin/python ${RUN}/share/trac/contrib/trac-post-commit-hook \
        -p "$TRAC_ENV" \
        -r "$YOUNGEST_REV"
fi

Emily, as Image Metrics have dubbed her, is based on actress Emily O’Brien. It’s not Emily O’Brien though. The producers asked O’Brien to pose for 35 facial images and then they went to work creating virtual Emily. New modelling technology allowed Image Metrics to control the most minute parts of her face to give Emily the most convincing movement we’ve ever seen.

Wow. Just wow. That’s so close it’s unreal, I was second guessing if it was real the whole way through. They did cheat by only rendering the face and not the hair, etc., but even so, that’s some impressive work. Now do it in realtime and put it in games!

All my scattered text files, del.icio.us bookmarks, blog entries, and overflowing file system folders just weren’t cutting it. I needed some way to take notes quickly, consolidate knowledge into a web-based respository that isn’t time-based, and organise my files with tags and folders.

All the local software (PersonalBrain, DevonThink, Journler, Tinderbox, Yojimbo, etc.) I tried didn’t fit. Tinderbox came close, but holy crap it’s expensive ($229?!). PersonalBrain did a great job of linking disparate information and linking files to notes, but it used up half of my Macbook ‘s CPU even when it was just sitting there, and it’s also expensive. All the others just had too many missing features or were too restrictive.

So instead of stuffing all my notes and files into one piece of proprietary software that might not be supported in the future, I’ve stuck to using open source apps. All my notes now get captured in xPad or on my public MediaWiki and all my files get tagged with QuickSilver and filed away in folders. It’s a quick, free solution that doesn’t suck computing resources on my laptop. The only thing missing is a high-level overview like Tinderbox’s map view, but I’m sure I can write a plug-in for mediawiki to do that one day.

How do you manage your personal information flood?

Download Resize ‘Em All.

Finally, a solution for my blog image resizing needs that doesn’t suck or cost me anything.

Open iTerm.

Go to Bookmarks > Manage Profiles

Choose Keyboard Profiles on the left and edit the Global Profile

Next to Mapping, click the + sign.

For Key, choose hex code.

In the text box next to hex code, enter 0×62 for b and 0×66 for f

For Modifier, check the Option Box
For Action, choose send escape sequence
Write b or f in the input field for hex code 0×62 and 0×66 respectively.

Now alt-f will jump forward a word and alt-b backwards a word.

Yay!

ViaVoice: Voice dictation for your Mac

With so much software getting developed, you’d think we’d get exactly what we need, but for some reason software developers don’t listen to customers as closely as they should.

What piece of software do you really need that doesn’t exist yet?

Maybe we can get some software developers to pay attention.

So I’ve put together my idea of a simple solution for you to critique. It may be a gap in the market, or it may be a stupid idea that has no chance of working in reality, which is why I’m kicking it into the wild so you can criticise it before I put too much effort into building it. Take a look at it and let me know what you think!

Summary: Find and network with people who have the skills you need, quickly and simply.

Problem:Too many complicated networking sites that lock you in and expect you to enter your life history and then demand payment when you want to do anything of value on the site. Overcharging for basic services and complicated sign ups mean that not many people want to use these sites, which reduces their value by at least 80% instantly, so the people actually paying for the service aren’t getting their money’s worth.

Solution: Simplified networking using just your name and keywords such as “Java, SQL, looking for contract, looking for permanent employment” where “looking for contract” and “looking for permanent employment” are common keywords that everyone will use so they don’t get offers they’re not interested in.

Once you’ve created your profile, you can search for a keyword, or combination of keywords, to find people to chat to. So if you’re looking for a Java contractor you’d search for “java, looking for contract” and then decide who you want to contact based on how their skills have been endorsed by others.

You can then leave a message in their inbox asking them to contact you, and they can reply or block you at their discretion.

Once they reply you can add them to your network, and they are requested to authorize the link. Either of you can delete the link if you want to.

When you’re networked with someone you can endorse their skills individually. For example, if you’ve worked with someone and know they’re good at project management, you can choose to endorse their skill with a short comment saying why you’re endorsing it. This pushes that person up in the rankings of “project management” searches.

If a certain number of people have blocked someone, then their profile will be flagged for investigation and probable black listing.

That’s the heart of the idea, other details will be fleshed out later once a workable prototype is online. Please let me know what you think and if you have any better ideas. Thanks!