We're very glad that Haiku has been part of Google's Summer of Code this year again. We were granted five student projects to improve Haiku. But since we had so many good and worthwhile project proposals, we set out to start our very own Haiku Code Drive. We asked for your help in the form of donations, and we were absolutely overwhelmed by the response we got from you, our community: we were able to sponsor 4 more student projects to work on Haiku.

Yes, of course, you know all of that already. The reason for this review is that, since both coding events are officially over by now, I wanted to give you an overview of what has happened, and how the students fared. Not all projects have been success stories, but we were lucky to have found some very talented students this year. We're glad we had you!

Google Summer of Code 2008

• Andrej Spielmann has implemented sub-pixel antialiased rendering in the app_server. He was probably the only student that was always ahead of his schedule, and could even deliver more than originally anticipated. He also easily adapted to our coding style and produced a lot of quality code.
• Dustin Howett intended to implement HPET support. At first, he struggled a bit with how to implement this in the best way, but eventually he found his way through this complex topic. However, he wasn't able to finish his project in time, and intends to keep working on it in the weeks to come.
• Zhao Shuai implemented swap file support in the kernel that has recently been enabled. While it would be fair to say that the FreeBSD implementation and his mentor, Ingo Weinhold, helped him out a lot, he was very receptive of critics and always willing to try to understand how the kernel internals are working. In the beginning, he also wrote some overview documents about our virtual memory subsystem.
• Alexandru Roman intended to add Zeroconf support to Haiku. However, he took a summer semester at school, and was surprised about the time commitment it asked for. He contacted his mentor, Ryan Leavengood, early on, and we're sad that he didn't manage to work on his project at all.
• Adrien Lemaire was supposed to write a CIFS client for Haiku to let it access Windows shares. However, he was a bit overstrained with the project despite his nice application. He also didn't find the time to dig into the project, and unfortunately didn't deliver anything.

Haiku Code Drive 2008

• Salvatore Benedetto intended to identify and fix most, if not all, of the remaining BFS bugs by first porting bonnie++ to Haiku. The plan was to add the missing functionality to Haiku that bonnie++ needs to run, in this case POSIX XSI semaphores. He experimented a lot with BFS, and we were able to fix several bugs together. He also implemented not only XSI semaphores, but also XSI message queues, and intends to complete his work by eventually adding support for XSI shared memory. During his project, he also ported the UDF file system to Haiku's current file system API. We're looking forward to see more from him :-)
• Jovan Ivankovic was supposed to port CUPS, or parts of CUPS and integrate them with the Haiku printing layer. Unfortunately, he was not able to work much on it due to his health situation. We wish him the best!
• Yin Qiu wanted to complete our ICMP handling in the networking stack. While he had a hard time with our coding style, he found his way through the stack, and came up with a good looking solution for error propagation and handling. Unfortunately it doesn't work yet okay, and is therefore not part of our repository yet, but he's continuing to work on his patch set. And we're patiently waiting for it :-)
• JiSheng Zhang has written a DV media node based on the Firewire stack he ported as part of last year's Google Summer of Code. As far as I understand, he couldn't really test his work yet, though, as his Firewire hardware got lost at Olympia. In any case, he intends to stay with us to improve his work in the future.

So while we had our failures, overall we're very happy with our students. We hope to refine our selection process for next year, if we're lucky enough to be part of the Summer of Code in 2009.

And since our own Haiku Code Drive was so successful this year (even if we saved a bit of the money), we plan to continue this project in the future as well. On behalf of Haiku, let me thank you, the donors, again for making this possible.

I would also like to thank our mentors that have devoted much time for their students, and guided them through their projects.

Thanks to Google Summer of Code student Zhao Shuai successfully finishing his project Haiku does now feature support for swapping. As of revision 27233 it is enabled by default, using a swap file twice the size of the accessible RAM. The swap file size can be changed (or swap support disabled) via the VirtualMemory preferences.

Swap support finally allows building Haiku in Haiku on a box with less than about 800 MB RAM, as long as as the swap file is large enough. I tested this on a Core 2 Duo 2.2 GHz with 256 MB RAM (artificially limited) and a 1.5 GB swap file. Building a standard Haiku image with two jam jobs (jam -j2) took about 34 minutes. This isn't particularly fast, but Haiku is not well optimized yet.

Haiku's swap implementation was heavily inspired by that of FreeBSD. At the moment it is not as sophisticated, but Zhao intends to borrow more of FreeBSD's optimizations.