I've spend a whole lot of the time of the last few days tinkering on various parts of my root server. With the time passing you get used to the comfort of various (web based) tools such as GMail, Google-Calendar, Google Reader, etc. You may notice that you just read the word "Google" quite often, so what pops into your mind? Right, privacy. Google kind of mines quite some of your data. Ever checked the Ads google shows you on various sites (given you don't use a capable Ad Blocker)? Sometimes it gets quite creepy. That data is quite valueable for profiling your behavior, and that profile is (not related to your persona, but in general) sold to marketing monkeys.
So, that's why you might want to rebuild all those Tools you're used to in a trusted environment - your own server. You don't want your mail stored in some possible hostile environment on a untrusted machine that could leak your valuable data. Turns out its not that easy sometimes. I've only worked on getting a capable Web Mailer and Feed Reader to run smoothly. What surprised me here - and why I'm writing about this - is that it was exceptionally hard - or rather time-intensive - for something sounding as easy as this.
I first targeted the Reader, looked around Freshmeat and SourceForge where you expect to find decent free software for that task. I've found quite a few not-so-simple-looking projects, including Tiny Tiny RSS. Turned out TinyTiny RSS is almost there, but needs PostgreSQL. So i set up PostgreSQL, installed TTRSS and set it up. Imported GoogleReaders OPML, and zup. worked. Problem's all the feeds are in one big Table - resulting in the whole thing beeing painfully slow. So up for 7-8 Hours of harcore Postgres performance tuning, trying to hack memcached into ttrss, etc. Speedup of almost 100%, yet it was not close to beeing usable. Turns out the developer didn't intent the project for archiving articles for having a searchable archive. So up for something different, Gregarius, which the TTRSS dev recommended. This worked quite out of the box, except for 3-4 Hours of tinkering and writing small plugins to get the whole thing to work properly. But it does - and has almost all the feature one expects.
The harder part came now, Webmail. I first tried to hack the roughly-set-up RoundCube that still was on my server. After short testing and many functions that just did not work due to unknown reasons i knew i needed something different - and started toying around with Horde and its webmailer Imp(4) in the Horde Webmail Edition pack. Horde feels somehow "unix style"-ish, like ... building a highly reusable backend, letting other projects include/work ontop of that backend, etc. - but let me say one thing: This beast is so darn hard to set up. I've got it working - more or less - after hours and hours of doc reading, tinkering around with mysql tables and databases and reading Horde source due to its ... slightly lacking ... documentation on some points. Still, it was so unstable and lacked features too. So i finally decided Horde/IMP was a bit too much to go with. After searching around its a lets-get-back-to-Roundcube.
Just that this didn't make things better, well... a bit at least. It takes tinkering, fixing old plugins to work with the current version, finding out why the hell buttons are greyed out that shouldn't be and getting a "well, reset the database (again)" from the developers. All of that fun. As of now i at least managed to get everything except for Filter-Rules and sa-learning ham to work. Phew!
Okay, so why am i writing all this? you may ask yourself. Well, i've gone through some PITA for beeing independent. It really takes some work to get everything to run smoothly if you're used to professional systems that are customly coded (such as the google stuff) and backed by real money its still some hackish tinkering to get the Free Software tools that we're given by the community (which i'm not ranting against by the way, all that code out there is really beautiful in fact) to the same level. I also want to encourage everyone out there not to give away their data but to build something on their own, keeping their data. As computer users used to for a good reason for a long long time.
You're going to use this stitch many times if you make plushies, usually used for closing after you've stuffed them. It's made of awesome. I plan to write a more detailed description this stitch in the future but these pictures have to suffice for now. You can google it in the meantime.
Like this (click for bigger picture). And pull every few stitches.
It'll close magically!