Mon, 17 Mar 2008

IPv6 continues to grow

Interesting report on how IPv4 was shut-off and everything continued to work.

Especially interesting was Google and the interesting logo effect.

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Tue, 12 Feb 2008

Hiss. Hiss.

Like most people I like good company. Good company and free booze is even better.

If you are in London, you missed out, as last week was when flag and bell was on. If you are going be around for their next event – 4th March 2008 – drop Megan a line.

Yes, people, it really is good company. And free beer.

Speaking of beer, those of you when a penchant for all things Python may want to come along to the London Python booze-up.

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Mon, 16 Jul 2007

Is the Google indexing all it is cracked up to be?

Like a lot of people, I've install Google Desktop for Linux.

First off, I've installed Beagle and Tracker before. So, I've played with a couple of these things. Obviously since I've installed GDL I felt that none of the existing Free Software solutions worked as well as I expected.

A while back I read Joe Shaw's note about how Beagle is tested with ~180K files. I considered this for a moment and realised that is why Beagle would turn my system into a pile of steaming dog poo.

My home directory has, normally, over 1,000,000 files (currently 893790 as I deleted some things before I fired up gtk-gnutella).

Considering Google handles web indexes significantly larger, you'd assume that their desktop search would be more than capable.

Alas.

I've had it installed for a week and it has achieved a 5% index. At least it does not chew up memory and only causes excessive CPU load. Oh, and the fact that it must have a browser (seemingly only Iceweasel will do) to use makes it even less useful. I'd let someone know but Google is also in on the whole "Let's make it as arduous as possible to report a problem" caper that seems to be common nowadays. Oh, well.

They went to all that time and effort to make a Debian package, why not put in the extra few minutes of work so that reportbug Just Works.

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Tue, 06 Mar 2007

postgresql forward / backward incompatibilities

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Fri, 24 Mar 2006

Some random odds and sods as well as some ACID

Horms, why not use gworldclock?

Martin, I think you are after tra.

Andrew, you do realise that you could do that with yum – which has been in Debian for a while now. You'd have probably heard about it earlier if Mako could be bothered to get Atom 1.0 support working on Planet Debian.

Rodney, Nokia phones generally use Netfront as their default. If you want to use Opera you have to specify it. I'd be highly skepticle of KHTML being used anywhere memory footprint mattered.

It looks like it might be WebCore (dervied from KHTML) instead. But that is only on very recent phones (first half of 2006).

Oh, and that is likely to be Yahoo merchant crawler. Yahoo probably thinks you are a corporate whore or something ;-).

Finally is it just me or does it strike anyone else as odd that the Mozilla Corporation, with all of its' money, is not able to get a crack group of hackers together so that Firefox 2.0 can render ACID2? At least two of the competitors can already do it, so, what's the go, Joe? (or in this case Mitchell).

[ / software] Trackbacks (0) Comments (1) permanent link permanent link

Thu, 16 Mar 2006

Back down on the (VCS) farm

Interesting how Sun have also come to the same conclusion[1] as I have with respect to Distributed SCMs. Only bazaar, git and mercurial matter.

Also interesting is that for immediate targets they also recommend Subversion. Maybe I've been channeling Bill Joy or something? Anyway, earlier I couldn't decide which of the three to use.

Now I know that git is the way forward. The (multiple) interfaces help enourmously (cogito, git, StGit, P(atchy)g, etc.). It means if you have some funky interface, you can migrate slowly by reimplementing what you use with git plumbing.

Just as Carl Worth talks about reimplementing Hg using git. Another major factor is the culture on the mailing list(s). List culture encourages/discourages other participants. And the git list culture encourages lots of people to share very small changes.

Which has meant that git has caught up very quickly and eclipsed everything out there[2]. It really is odd reading that email and seeing just how far git has come.

Anyway it looks like what Sun are considering doing is that they might reimplement the interface of their internal tool (teamware) on top of git. If they can make that available, it'll advance things for everyone (again).

1. http://mail.opensolaris.org/pipermail/program-team/2006-March/000172.html
2. http://mail.off.net/pipermail/codeville-devel/2005-April/000013.html

[ / software / vcs] Trackbacks (0) Comments (0) permanent link permanent link

Tue, 14 Mar 2006

Here a VoIP, there a VoIP, everywhere a VoIP.

I've been using VoIP for many years. Some thoughts.

Asterisk has never really appealed to me. I can't put a definite finger on why this is so but it feels like having email printed. A half-step along the way. It is a good halfway point but doesn't feel like a full solution.

What does is something which turns PSTN into SIP; then you can use a SIP router (SER, OpenSER, partysip, sipXrouter and many others) to do what you need.

That kind of device is something like a Sipura 3000. Very easy to setup to register to your own SIP router, where you can do the heavy lifting. Of course, you don't need to interface to the PSTN to take advantage of VoIP.

I have SIP accounts at bbpglobal, ekiga, fwd, iptel, openwengo, sipme, sipgate and voipuser. As well as few other, more private, networks.

NOTE: openwengo requires some work to retrieve the SIP id.

I've also been playing with gaim2 recently. It is the reason I'm only vaguely online – gaim2 breaks sound on my laptop – and that interferes with the VoIP stuff I do.

Anyway, one interesting feature of gaim2 is that it supports SIP IM extensions (presence basically). I was trying to debug why this wasn't working which meant I was doing some tcpdumping and noticed some SIP messages go past — I normally have twinkle (currently the best SIP softphone on Linux) running, so this wasn't unusual.

What was, is that the originating machine wasn't my own. It was my mothers computer. I decided to investigate.

It turns out that Yahoo Messenger (which my mother has signed up for so she could speak to her nephew) can actually speak SIP. This could also be part of the impending MSN / Yahoo IM networks interop. that is being completed.

I'm not sure how they will achieve that but from looking through the packet traces of Yahoo's "PC-to-PC" calling, some kind of negociation takes place of the YMessenger channel and a SIP channel is established between the client and a seperate set of (Akamai-sed) media servers. It also seems like they have some kind of SIP proxy which normal/old clients can speak to and which speaks back SIP.

From the network traces I've done, one side sends a cookie down the YMessenger protocol pipe, and the other side responds via the SIP pipe. From further traces no Linux client can (currently) interoperate with Yahoo's service until the client supports SIP over TCP, SIPS, RTP and RTCP, not to mention Yahoo's special cookie scheme.

Now both networks, Yahoo and MSN need to open themselves up so that they provide presence information externally (via SIP) and allow users to talk to anyone with a SIP address anywhere. You know, just like you can with email.

[ / software] Trackbacks (0) Comments (1) permanent link permanent link

Tue, 17 Jan 2006

Old McDonald had a version control system …

The great thing about standards is that there are so many to choose from

The same, of course, goes for VCS. I've used RCS, CVS, Subversion (svn), Arch (larch, tla and baz), Mercurial (hg), git, Bazaar-NG (bzr), svk and no doubt a few others I've forgotten, or wished I had.

The current state of the art in Free Software is Subversion IF you have to interoperate with developers who use Windows. That's it. Don't read further if you have Windows-based colleagues.

If, though, you get to choose the technology, then based on my experience with all of the above you then you should pick between: git, hg and/or bzr.

None of RCS, CVS nor svn are useful in the long-term since they do not allow for distributed development and branching. Yes, it can be done — a good example is svk — but not without difficulty and lots of gnashing of teeth. In svk's case, you need to be careful when merging since svn (which is what svk uses underneath) does not keep history or context.

I hit onto Arch and was lucky enough to stumble upon Robert Collins who guided my understanding of it and changesets. Unfortunately Arch was led by Tom Lord, who has real issues with generating a community of people around him. His first effort (larch), stagnated was re-written (by Robert) in C++, and called barch, while he wrote tla.

tla development also stagnated and, eventually, Canonical forked it and called it baz. I was very skeptical about a company like Canonical sticking to their promise ("we'll develop it until version 1.5 where it should be a good transition point to our next system"). I wasn't surprised when they renegged on that committment. It is their right of course, but it is a shame. Right now, if you have to use Arch — and you haven't used it before — use tla. If you have run into the limitations of tla, then use baz but bear in mind that baz is stone-cold dead as far as development is concerned. tla, is only cold dead.

Which brings us to bzr, hg and git. All three of these systems are very similiar.
Distributed development? Check.
Portable to everything (but Windows, and that is changing too)? Check.
Implement the same generic object-based filesystem? Check.

I've stayed away from bzr since Canonical is supporting it; that isn't bad but it does mean it'll forever be tied to what they want. It also means the community around it is extremely small. bzr is interesting in that it classifies everything as an object (directory, file, etc.) and stores them in pre-classified directories. In particular it stores file data in a 'weave' format (similiar to SCCS).

hg has a small, vibrant community around it but the problem is the community is small. And, like bzr, it also stores object in pre-classified direcoties – although in a delta-compressed format for space reasons. Both are written in Python (more about that latter) and have nice graphical tools to do things like quilt and three-way merges as well as visualising development history.

In the end I've chosen git as my future revision system; it has a development leader who encourages rather than discourages. The documentation overshadows everything except CVS and svns – and is on par with them. It is fast. it has a large community behind it. You can use it at a number of levels (git,cogito, stgit, etc.).

It has great support for importing to everything except bzr and I'm yet to find any particular issue with it. Plus it merging is very good. In fact, as Linus points out you can have any merging startegy you like. So, any theoretical advantages of hg and bzr are just that, theoretical.

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Old McDonald had a version control system …

The great thing about standards is that there are so many to choose from

The same, of course, goes for VCS. I've used RCS, CVS, Subversion (svn), Arch (larch, tla and baz), Mercurial (hg), git, Bazaar-NG (bzr), svk and no doubt a few others I've forgotten, or wished I had.

The current state of the art in Free Software is Subversion IF you have to interoperate with developers who use Windows. That's it. Don't read further if you have Windows-based colleagues.

If, though, you get to choose the technology, then based on my experience with all of the above you then you should pick between: git, hg and/or bzr.

None of RCS, CVS nor svn are useful in the long-term since they do not allow for distributed development and branching. Yes, it can be done — a good example is svk — but not without difficulty and lots of gnashing of teeth. In svk's case, you need to be careful when merging since svn (which is what svk uses underneath) does not keep history or context.

I hit onto Arch and was lucky enough to stumble upon Robert Collins who guided my understanding of it and changesets. Unfortunately Arch was led by Tom Lord, who has real issues with generating a community of people around him. His first effort (larch), stagnated was re-written (by Robert) in C++, and called barch, while he wrote tla.

tla development also stagnated and, eventually, Canonical forked it and called it baz. I was very skeptical about a company like Canonical sticking to their promise ("we'll develop it until version 1.5 where it should be a good transition point to our next system"). I wasn't surprised when they renegged on that committment. It is their right of course, but it is a shame. Right now, if you have to use Arch — and you haven't used it before — use tla. If you have run into the limitations of tla, then use baz but bear in mind that baz is stone-cold dead as far as development is concerned. tla, is only cold dead.

Which brings us to bzr, hg and git. All three of these systems are very similiar.
Distributed development? Check.
Portable to everything (but Windows, and that is changing too)? Check.
Implement the same generic object-based filesystem? Check.

I've stayed away from bzr since Canonical is supporting it; that isn't bad but it does mean it'll forever be tied to what they want. It also means the community around it is extremely small. bzr is interesting in that it classifies everything as an object (directory, file, etc.) and stores them in pre-classified directories. In particular it stores file data in a 'weave' format (similiar to SCCS).

hg has a small, vibrant community around it but the problem is the community is small. And, like bzr, it also stores object in pre-classified direcoties – although in a delta-compressed format for space reasons. Both are written in Python (more about that latter) and have nice graphical tools to do things like quilt and three-way merges as well as visualising development history.

In the end I've chosen git as my future revision system; it has a development leader who encourages rather than discourages. The documentation overshadows everything except CVS and svns – and is on par with them. It is fast. it has a large community behind it. You can use it at a number of levels (git,cogito, stgit, etc.).

It has great support for importing to everything except bzr and I'm yet to find any particular issue with it. Plus it merging is very good. In fact, as Linus points out you can have any merging startegy you like. So, any theoretical advantages of hg and bzr are just that, theoretical.

[ / software / vcs] Trackbacks (0) Comments (0) permanent link permanent link

Wed, 07 Dec 2005

dead:cafe::6

Or, how to get thyself onto IPv6. A lot of people when I mention IPv6 say Yeah, but who else is using it?. As of last year (2004) there are roughly 200 million Windows XP (or better) machines with IPv6 defaulted on. [1] [2]

That doesn't include Linux, *BSDs, etc. If all these people have got IPv6 (oh, that is just those without NAT) then you've got to wonder just how many people in total have it enabled.

The next question I commonly hear is Yeah, but what can you do with it?. Of course you can run things as you were before and that's fine.

Actually it is more than fine. Here is me pinging my default gateway.

IPv4

eve:[~]% sudo ping -c 1000 -f 203.7.227.129
PING 203.7.227.129 (203.7.227.129) 56(84) bytes of data.

--- 203.7.227.129 ping statistics ---
1000 packets transmitted, 1000 received, 0% packet loss, time 415ms
rtt min/avg/max/mdev = 0.209/0.237/0.720/0.076 ms, ipg/ewma 0.416/0.564 ms

IPv6

eve:[~]% sudo ping6 -c 1000 -f 2001:388:c0c5::1
PING 2001:388:c0c5::1(2001:388:c0c5::1) 56 data bytes

--- 2001:388:c0c5::1 ping statistics ---
1000 packets transmitted, 1000 received, 0% packet loss, time 263ms
rtt min/avg/max/mdev = 0.174/0.195/0.577/0.016 ms, ipg/ewma 0.263/0.196 ms

IPv6 is faster to process on the same hardware. Gentoo ricers take note! Okay, so you can run all your same protocols on top (and get a speed benefit) but what else?

Well things that require peer-to-peer connectivity, like VoIP and filesharing can particularly benefit. For example, gtk-gnutella recently acquired IPv6 support. And most 3G phones also use IPv6.

At the moment, though, most people are expending effort actually getting connected. Within Australia there are two way

• the AARNet tunnel broker at http://broker.aarnet.net.au
• using 6to4 (RFC) addressing.

You can use either mechanism but if you have a dynamic IPv4 address the tunnel broker will be a better option as it will mean your IPv6 addresses will be fixed. If you are using Debian you will need to install the tspc package – let me know of any problems as I am the maintainer.

If you have a fixed IPv4 address you can use 6to4. 6to4 converts your IPv4 address (which is 32 bits) into hexadecimal and then adds '2002' to the front. Here is a quick example:

IPv4 address: 138.25.6.65 (decimal)
              8a.19.06.41 (hexadecimal)
IPv6 address: 2002:8a19:0641::/48

You would then set your default router to be the 6to4 anycast address of either 192.88.99.1 (IPv4) or 2002:c058:6301:: (IPv6) — 10 points if you worked out the IPv6 address in advance.. That will get you to the closest topological point of interconnection between you (IPv4 mongrol that you are) the pristine shinyness of IPv6. In theory.

In practise, there are closer points at AARNet (ipv6.aarnet.net.au, 192.231.212.5) and at Telstra (203.14.5.7) but they are not advertised by the routing system. Which is why I currently recommend that even with a static address you use the tunnel broker — it's faster.

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Thu, 24 Nov 2005

All webbrowsers suck...

eve:[~]% firefox
*** glibc detected *** free(): invalid next size (fast): 0x08bc9228 ***

eve:[~]% amaya
04:22:56: Deleted stale lock file '/home/anand/amaya-anand'.
*** glibc detected *** free(): invalid pointer: 0x0a47bb30 ***
*** Amaya: Irrecoverable error ***zsh: abort (core dumped)  amaya
eve:[~]% galeon
*** glibc detected *** malloc(): memory corruption: 0x081dcbf8 ***

eve:[~]% epiphany
*** glibc detected *** free(): invalid next size (fast): 0x08561540 ***

Fortunately I can still browse the web within Emacs but not Vi (or vim). For anyone else experiencing the same problem, MALLOC_CHECK_=1 might allow you to run some programs further (doesn't work for me though).

Update: It turns out my problems were related to SCIM and an upgrade solved my issues. One things I did learn is that firefox --debugger=gdb will give you a chance to get a decent backtrace, especially if you have the relevant -dbg libraries installed.

[ / software] Trackbacks (0) Comments (0) permanent link permanent link

Thu, 10 Nov 2005

Plugins I have installed

Michael Still writes about Trackback pings.

I'm still not used to blosxom and how everything fits together but I thought I'd enumerate the plugins I've installed.

categorytree

I've actually got this named as 01categorytree, so it runs first, but it gives me the listing of topics I've posted too

extrensionless

I'm a big fan of Cool URIs, this helps me almost get there. Bonus points to those who can work out where I don't quite make the magic happen seemlessly. Oh, named 02extensionless.

atomfeed

As every good blogger knows, syndication is the name of the game. While you got use the deprecated RSS, Atom is where is it at. I've actually disabled RSS feeds in preference to atom

breadcrumbs

On permalink posts, you'll see a click-trail. All professional websites have them, so I figured I should too.

calendar

Presents a calendar with all the posts you've done on various days highlighted, so you can quickly determine when someone has been sleeping or has been an insomniac

config

Comes with the Debian blosxom package, and originates with the author. Allows you to customise configuration dependant on where the post is in your heirarchy and other magic.

directorybrowse

If you have deep category hierarchies this makes all the elements of the paths clickable. It's cool, you want it.

feedback

Part of the blogging scene is to allow other people to comment upon what you've written. This gives them comments and trackbacks. I've modified this somewhat and still need to fixup how comments are previewed but otherwise this plugin is gold

flavourdir

Also part of the blosxom package, this means you do not have to have your blog entries intermingled with the page templates (known as flavours) used to display them

interpolate_fancy

For large plugins, like feedback, you might end with some text on your template that you only want displayed some of the time. This enhances blosxom so your template (flavour) can have conditionals, comparisons and whatnot.

multcat

Not everything can be easily catagorised into one thing. This lets you put a post into multiple places. It'd be nice if all the categories a post was under was displayed as part of the category path

rendertime

Calculates the amount of time the page took to render

On my todo list is to convert from a table based layout to a fluid CSS one. It'll probably be a 3-columned look with fixed-width sidebars since negative margins looks are harder to get right.

[ / software / blosxom] Trackbacks (0) Comments (0) permanent link permanent link

Wed, 09 Nov 2005

Blogging with a phone

Or 'mobloging' for short. Basically it is using your mobile phone to add content to your blog. All very interesting, except when you get to the details.

Stewart Smith gave it a whirl ([1], [2]) with fairly predictable results. This is because the message you might receive is dependent on the phone manufacturer and the MMC does to your message. And that is before it might be mangled along the way via email.

Basically most phones (or MMCs) will convert the content to either Base32 or, more commonly, Base64. Generally the MMS is composed of only two parts, and then sent to an email address. From there you'll have to write something that takes the email and converts it into a blog entry.

A while ago a friend of mine wanted to do the same, so I wrote some perl to do it for him. It seemed to work for him (with some tweaks) and it might be enough for you too. You can grab it here However you might find it simpler to use GNU Arch and register http://www.progsoc.org/~wildfire/arch/projects/ and then browse in wildfire@progsoc.org--projects/moblog.

[ / software / moblog] Trackbacks (0) Comments (0) permanent link permanent link

Thu, 03 Nov 2005

IPv4 Link Local addresses

James Henstridge notes that zeroconf can cause large amounts of network traffic. Yes, that was true back in July. A bug was filed and then fixed within 2 days.

This was because I was at DebConf and able to see how zeroconf performed in a large environment. Basically there are two tests you do to determine whether an ARP packet is destined for you. And at the time, zeroconf, was only doing one of them.

Davyd asks why would you use zeroconf over NetworkManager. I haven't been able to put my finger on exactly why, but I feel the design is wrong.

I do believe there should be an overall co-ordinating process looking after networking on the user-space side of the fence though. I've been thinking about writing something which would do it too; but I haven't fleshed out my ideas enough to commit my editor to coding.

Oh, one bug that NetworkManager does have is that it only allocates IPv4LL addresses when it fails to acquire one from the DHCP server. This is wrong and goes against RFC3927.

[ / software / zeroconf] Trackbacks (0) Comments (0) permanent link permanent link

Wed, 02 Nov 2005

Yay!

First post. I too have decided to get into the whole blogging scene.

I looked around at various pieces of software and only blosxom and pyblosxom fitted my criteria.

My criteria was pretty simple:

• Plenty of features (comments, calendar, categories)
• No database (even though I prefer Postgresql, I didn't want to have to learn a schema if I decided to migrate or it all went cactus
• Vibrant community

Both blosxom and pyblosxom have those but I found it easier to get blosxom setup and running. Eventually I'll write up a colophon about what I have and how it is configured. However, please excuse the very rudimentary setup I have — it'll get better in time.

[ / software / blosxom] Trackbacks (0) Comments (0) permanent link permanent link