Today Little Go 0.7 has finally been released on the App Store. Although the app was initially rejected by Apple's review team, I was quickly able to point out why the reason given was clearly wrong. Although I am happy that the app is in the store now, I am still a bit peeved at how arbitrary the rejection seemed to be - it almost looked as if the reviewer hadn't even bothered to launch the app a single time. Fixing a genuine bug would have been OK, but seeing 10 months of work being casually dismissed with a copy & pasted statement is a bit hard to swallow...

Anyway, the 0.7 release of Little Go adds no new features, but fixes a couple of serious bugs. Other than that, the main work that has been done is the addition of artwork such as an application icon, and in-game icons for the tab bar and the toolbar on the game view. As usual you can grab the sources from the project page.

Despite the original plan to focus on stability, a few things that could also be seen as features have managed to creep into the new beta release of Little Go. Of course, if you ask me I will say that those feature-like things in reality are bugfixes ☺ that were necessary to make Little Go ready for the App Store. Care for an example? A stone is now displayed immediately when you touch the board with your fingertip - a great improvement in my opinion! Also the mess with GTP engine settings has been finally cleared up, and... wait, why am I writing this? Check out the project page and see for yourself...

The new beta build 0.5 of Little Go adds handicap and komi selection to the "New game" view, and even more importantly, there is now a proper scoring feature available that calculates an accurate territory score and lets you mark stone groups as dead or alive.

I have decided that the app is now feature complete for the first public release to the App Store. In the next iteration the focus therefore will be on fixing the most glaring issues and running the app through Instruments. I have never used Instruments before, so this will be an interesting experience. Hopefully not too interesting...

Ad-hoc distribution is definitely the way to go to make it easy for beta testers to install an iOS app that is not yet available through Apple's App Store. The answer to this stackoverflow question thoroughly explains how it works.

Today I tried to create a new ad-hoc distribution for my Little Go app, the first after iOS 5 has been released. Very much unexpectedly, the whole process fell flat on its nose due to a combination of code signature and entitlement problems. After a 4 hour battle I now have things working again, and although I am not entirely sure about the exact source of the problem, I thought I would like to write down what I have learned this afternoon.

The final solution, anyway, for those who don't want to read the whole story: Invalidate and then regenerate all provisioning profiles on Apple's iOS Provisioning Portal.

After many months of slow but steady development my newest pet project Little Go has now reached a state where I feel confident enough to post a release note on the front page of herzbube.ch. Little Go is an iPhone app that lets you play the game of Go on your favourite smartphone or on your iPad. You can play against another human, or against the computer.

Little Go is still under development and not yet available in Apple's app store, but you can contact me if for some strange reason :-) you are interested in beta-testing. Please refer to the project page for details about which information I need from you, or if you simply want to download the tar ball and look at the source code.

In September 2009 I started to use greylisting after the spam rate on my main email address had reached an unbelievable 1200 messages per day. Here's the article that describes the event, the reasons why I avoided greylisting for a long time, and why I finally had to give up my resistance.

It is now 15 months later, and I just added an entry to my spam statistics page which makes me a very happy email user! The spam deluge has been reduced to a trickle of merely 36 messages per day, of which almost 95% are correctly recognized as spam by SpamAssassin. If ever there was a success story, greylisting at herzbube.ch is it!

After the break there is a little timeline that might be interesting to read. It is essentially a summary from the statistics page linked above.

Last week I tackled another upgrade of the Bugzilla site hosted on my Debian server. Unfortunately, I found out the hard way that the package that I tried to install (bugzilla3-3.4.7.0-2) was badly broken - I was left with a half-completed installation that simply would not work. I reported the problem but got no response, so I decided to get to the bottom of this and find out what the trouble is. This is not the story of what I found out, but of how I was finally able to fix the problem.

The main new thing in version 0.4 is that mkroesti is now available as a CGI script. Check out the demo site if you are interested. The demo site is no longer operational. Visit the project page for more information and to download the tar ball.

Two good news from the Mediawiki battle front of a "grizzled Debian veteran" :-) Read the whole article if you are interested.

Today I found out that there are simpler things than configuring an Apache virtual host for Bugzilla to live in. With a bit of URL rewriting magic, the basics are moderately difficult to solve. What drove me crazy, though, was that the CGI scripts stubbornly generated links to /bugzilla3/skin/[...], instead of just /skin/[...].