Quite a number of people are aware of the PAE which can extend the addressable space from 32bit up to 36-48-52bit (depending on the implementation; as I understand, Windows PAE extends to 36 bits, or 64GB of addressable space). However, overwhelming numbers of internet pages continue insisting that a not-more-than-4GB limit for the 32bit Windows is the consequence of 2^32 = 4GB architectural limit.

There is an excellent, in-depth, well-argumented article by Geoff Chappell on the issue. Highly recommended in its entirety to those who want a complete understanding (additional side-reading and facts verification might be necessary).

A single citation to get you started:

There is already on the Internet and elsewhere an awful lot of rubbish to read about this question. Hardly any of it would be worth citing even if I didn’t want to spare the authors the embarrassment. A surprising number of people who claim some sort of attention as expert commentators would have you believe that using more than 4GB of memory is mathematically impossible for any 32-bit operating system because 2 to the power of 32 is 4G and a 32-bit register can’t form an address above 4GB. If nothing else, these experts don’t know enough history: 2 to the 16 is only 64K and yet the wealth of Microsoft is founded on a 16-bit operating system that from its very first version was designed to use 640KB of RAM plus other memory in a physical address space of 1MB. Some remember this history and add seemingly plausible qualifications that exceeding 4GB is possible only at the price of nasty hacks that require everyone—well, all programmers—to jump through hoops. Fortunately, Intel’s processors are a lot more advanced than the 8086 from all those years ago.

P.S. Unfortunately, patching the kernel won’t help make Windows XP see more than 4GB RAM: even though the kernel itself does support more RAM (with PAE), starting with SP2 the HAL was modified in a way prohibiting access to any RAM beyond 4GB. Patching may only be suggested to devoted geeks with Vista’s and 7′s.

rtorrent is an excellent ncurses rtorrent client.

There are a few highly-popular patches, which haven’t yet made it into the Squeeze’s rtorrent version 0.8.6.

Note: ArchLinux already has an rtorrent-extended package, so you may want to look at that one (instead of my Debian package below). I might use Arch’s patchset next time to generate a truly extended rtorrent package.

I’ve incorporated 3 of such patches (#239 ipfilter, #2064 GeoIP support, and #2171 colours/eyecandy), and built a squeeze amd64 rtorrent-enhanced package.

A few excerpts to explain how these extra features are to be configured and used (all come from the above-mentioned trac tickets, some were edited/extended).

ipfilter

ipfilter allows to selectively blacklist/whitelist peers – based on IP address range files (so-called ipfilter files). The ipfilter files may, for example, come from bluetack.co.uk. Each line of the file contains a record in this format: range_description:start_IP-end_IP, where start_IP should be less than or equal to end_IP. Multiple files can be used. Overlapping ranges are merged automatically. Both incoming and outgoing connections are checked against the filter. Exclusions are not supported, so connection to/from IPs in all the loaded ranges will be disallowed and dropped.

Include “ip_filter=” directive in .rtorrent.rc. For example (paths are specified relative to user’s home directory):
ip_filter=ipfilter/level1,ipfilter/level2

It is probably a good idea to reload ipfilter files once in a while, so you can also include “reload_ip_filter” directive on schedule to refresh the filter. The same files named in “ip_filter” will be reloaded.

schedule = filter,18:30:00,24:00:00,reload_ip_filter=

To support the feature you may want to setup cron job to reload and unzip files from bluetack. Examples here and here.

GeoIP

In the Peers view of each torrent a new column – CC, country code – is added, which shows peer’s country code. When examining each peer, you will see more data from the GeoIP database (if you have the relevant files installed): AS num and city. At the very least, you need to have geoip-database installed. You can get free “lite” versions of city/AS-databases, rename them (removing “lite”) and put into an appropriate location to make rtorrent+geoip use them as well. Geop-isp data support isn’t integrated into this rtorrent package.

Colours

This patch somewhat changes the way rtorrent shows torrents. Seeding torrents have bold titles, there are no half-displayed torrents at the top/bottom when scrolling, and you can configure colours for active/done torrents (ticket libtorrent.rakshasa.no/ticket/1382, which seems to be gone now). Source code and testing hint at these configurable colours: done_fg_color, done_bg_color, active_fg_color, active_bg_color, and at these possible values: 1 (red), 2 (green), 3 (yellow).

Geek’s cellar
A related (though not used in any way for the preparation of the package) resource is rtorrent mods page.

Relatively schematically, applying patches and building the package was performed in these steps (starting within some newly-created directory):

1. sudo aptitude install cdbs devscripts [and whatever else you find you're missing]
2. apt-get source rtorrent
3. dpkg-source -x rtorrent_0.8.6-1.dsc
4. cd rtorrent-0.8.6
5. cdbs-edit-patch 01-ipfilter.patch
6. patch -p1 < /path/to/patch/239
7. exit 0
8. cdbs-edit-patch 02-geoip.patch
9. patch -p1 < /path/to/patch/2064-after-ipfilter
10. exit 0
11. cdbs-edit-patch 03-eyecandy.patch
12. patch -p1 < /path/to/patch/2171-mod-with-canvas
13. exit 0 [you could do all 3 patches together, but I prefer cleaner and manageable approaches]
14. [some weird operation to be explained below]
15. dpkg-buildpackage

Weird operation: I didn’t know how to make dpkg-buildpackage run autorun.sh (which is required for one of the patches I’ve used) before configure (could someone enlighten me how to do that?), so I started a new patch with cdbs-edit-patch, then ran one by one `autoreconf -if`, `./autorun.sh`, `./configure`, exited with zero status and was done with that problem. The resulting “patch” was 2MB large

This overview presentation is two years old, but still a highly valuable resource: modules and tools mentioned are alive and useful.
I think this is the second presentation by Giovanni I’m embedding (first one being about GNU/make for bioinformatics).

Imagine you need to install pycassa (which uses easy_install). Here are the 2 (at maximum) very simple steps to have it properly debianized and installed on your Debian/Ubuntu:

• if you don’t have the python-stdeb package: sudo aptitude install python-stdeb
• pypi-install pycassa

That’s it.

Refer to stdeb readme for more information. You will need that if there are dependencies – which might not be resolved automatically by stdeb.

Before stdeb, it wasn’t exactly trivial to make a .deb from python module.

HandlerSocket provides a direct access to InnoDB storage, bypassing SQL interpretation layer. With in-RAM data, it may raise MySQL performance to 750000 queries per second.

Updated on the 5th of March, 2010 (added flowdock and pivotal tracker, and also personal experience using a few of the previously described tools).

Back in 2007 I wrote a brief review of web-based project management tools. After that, I started using dotProject for personal projects management. I’m still using it, but for collaborative project management, communication, and tasks/milestones tracking dotProject isn’t perfect.

I need a tool, which is

• collaborative
• web-based (to allow effective collaboration)
• preferably free
• has concise per-project activity log
• minimal required functionality: tasks, milestones, files, and status updates.

After trying a few things, our small team settled for now on using github + pivotaltracker jira + confluence + flowdock.

Here’s a full list of tools briefly reviewed. I’ve been already using ProjectPier, so I’ll start with this software.
Read the rest of this entry »

On the 24th of September 2010 Microsoft posted a video showcase titled “A few perspectives on OpenOffice.org”. Here’s the page with the video: http://www.microsoft.com/showcase/en/US/details/faaf9eb8-77c6-4bed-bc08-c069a7bfbb04. It asks to install silverlight, and if you don’t want that – look for the Watch as WMV direct videostream link.

Just a single quote from Glyn Moody, Computerworld UK:

The criticisms made in the video are not really the point – they are mostly about OpenOffice.org not being a 100% clone of Microsoft Office, and compatibility problems with Microsoft’s proprietary formats. The key issue is the exactly the same as it was for the Mindcraft benchmarks. You don’t compare a rival’s product with your own if it is not comparable. And you don’t make this kind of attack video unless you are really, really worried about the growing success of a competitor.

See also what Savio Rodriguez (Infoworld) has to say about that video.