I've written an evil bash script that will (just about) resolve dependencies of an rpm based distributions files, and install them in to a chroot. It appears to work on my debian system quite well, with just the straight debian version of "rpm" from unstable, and some sed, grep and temporary files.

If you're interested in taking a look, then it's available at http://www.sommitrealweird.co.uk/development/rpmbased-dist-chroot.html .

Why doesn't Win2k3 have a nice way of exporting "Local User and Groups" for importing to a different Win2k3 server? At the moment, this would be really handy to be able to do. Oh well, I'll have to see if doing a restore and not letting it reboot and then restoring the hardware configuration works.

Not long left now, though... only 9 more working days for Paston.

In 2 and a half weeks, I'll be moving to Brighton... So, here comes the fun bit... Leaving Drinks type Evening of doom needs planned... it'll need to be Thursday 15th September as I'll be driving places on Friday, then more driving on Saturday, then more driving and Trains on Sunday... I could easily be convinced of going out for drinks at weekends other than that, but the Thursday is the day...

So, to that end, if anyone fancies coming for a leaving drink on Thursday 15th Sept, I'll be in the Fat Cat from ~ 1930 (that's 7.30pm for those that never understood the 24 hour clock), please feel free to join me :)

In other news... Brighton beware!

Well, today, my first day back at work after a one week 'holiday' (which included a nice job interview with scary scary people, but then I'm scared of people), is over.... The letter of resignation was handed in this morning (woo!), and I shall only be at Paston Chase for another 4 weeks (ending on the 16th Sept).

I'll be leaving Norwich and moving to the insanely strange place called Brighton. I might post further details later, if I get round to it.

I've added in some support for setting important and critical levels on a per maildir level in my statusd_maildir.lua script. Means that if you have some less important or more important maildirs you can make them go pretty colours quicker.

Seems to be working reliably for me... now if I could work out how to make the statusbar bigger (like, 2 lines high rather than one, and with the mailboxes all on the bottom line and the top line for date, time, load, cpufreq, etc etc) it'd be very handy... anyone got any hints? I glanced at the statusbar code but I haven't yet figured it out, and it doesn't appear to automatically wrap if it over fills, which is a PITA.

I've switched to using Ion3 as my window manager, it's a nice idea, and I'm getting kinda used to using it, doesn't half change perspective on using graphical applications though. Ion3 has a rather nice single line statusbar by default that will show the time, load and mailcount for your MAIL environment, assuming that you're using mbox mailboxes... Now, I don't use mbox for mail anymore, I offlineimap my mail in to ~/Maildir/INBOX ~/Maildir/project-list etc etc, so, first task... lets see if there's a Maildir mail monitor... looking around provided no useful results, so, I wrote one. First attempt at diving in to lua (though, this code has now been looked over by rjek so, it should at least work for more people than just me), the code is available from my ion3 script repository, it's the only script there at the moment, but I'm sure that other things will bother me sometime and I'll write some more.

All comments gratefully received :)

You may remember that about a month ago, I posted about some thunderbugs getting in to my laptop screen, I phoned Acer and got assigned a number and told to phone City Link to arrange collection... I finally got round to phoning City Link on Monday, the laptop was collected on Tuesday, and I'm now typing this on it... with a bug free screen!

Other than the first time I sent the laptop back, the rate at which they have fixed it and returned it has been very prompt. The difference between the first time and the other times? Simple, I had paid for an Acer Advantage warranty when I bought the laptop, I failed to state this the first time I returned it, and thusly I waited a month to get the laptop back...

So, if you want a speedy service, and no courier charges... with a quick turn around... Buy Acer and the Acer Advantage warrenty.

Anyways - on to the chips that I bought on the way home!

Argh, my laptop screen is being attacked by evil black things, there's now 2 of them in the screen, it's not dead pixels as earlier on I was watching as one roamed about, I think that they are both now dead, but they're on the inside of the screen as far as I can tell. I have a shiny new CaseID from Acer and will be sending the laptop back on Monday to get them removed, hopefully without them charging me for the screen, as technically there's nothing wrong with it, other than 2 evil little black things roaming about.

Yesterday our leased line from UUNet (MCI) was cut off by BT at around 1.30pm, 7 days early. It should have been cut off next week on the 30th, as such a lot of rushing around, praying and general chaos has taken over while we continue (with the machines now in this building on the end of an ADSL line, and not accessable to the outside world) to move anything remaining off of the servers.

Of course, that was also our primary DNS server, yay for BT :(

First, a big congratulations to all that were involved in the debian release process, you've done a fantastic job. It's a shame that the CD images weren't perfect first time round, but it's fantastic that it was picked up so quickly and publicised to avoid people going manic on creating CDs. Thank you all for your fine work, and remind me if I meet you guys to buy you a beer or two :)

Now on to other stuff... I know that xmms-scrobbler didn't make it in to Sarge, I haven't yet had time to fix the important bugs that are filed against it, if you're one of the bug submitters, and I haven't responded to you yet, I'm very sorry, it's been a manic couple of months at work, and it's not going to slow down for the next month or so. I need to go through a fair chunk of the code, especially the tag reading code, to stop it from segfaulting, and I think I've spotted where the issue is with the submissions of queued tracks, I'm not impressed, currently, by the queueing code, an assumption that I'd made on how it works appears to be untrue, so I'll try to look at that in the next couple of weeks if I get some breathing space and don't waste it in relaxing.

And the other thing... I'm starting to think about working on another onak backend, basically a caching backend so that you query a local server and if it doesn't know about a key, or hasn't got that key recently (in a configurable amount of time), it'll go and grab it again. I've also noticed that sometimes the keyservers out there aren't quite in sync with each other, so the other idea for the backend is to get the results from more than one upstream keyserver, merge them in the server, and if the key we end up with is different to the key got from a particular server, we send back up the key we've now got. Can anyone see anything wrong with this approach? If not, I'll start coding that some time this month (hopefully).

I think that's all for now, I'm currently fuelled by a bottle of cheap merlot, so this post may make no sense, but I'm fairly relaxed ;)

Cheers,

Brett.

As quinophex runs Ubuntu Hoary, and I happen to have a pbuilder set up to build hoary packages on my slaptip, I've compiled up freeciv2 packages for hoary from the current debian sources (v2.0.1). To grab them add:

deb http://www.sommitrealweird.co.uk/ubuntu/ hoary freeciv2

to your /etc/apt/sources.list file. They're straight off compilations of the debian sources in a hoary pbuilder, and it all compiled cleanly, and it was working for the quinophex.

Thought I'd blog it just incase anyone else was interested (it's insanely addictive).

Thanks to Noodles, Dave, Becca and John for coming out for some birthday drinks last night. Many thanks for the shiny present from Mr Noodles. Maybe soon my head will forgive me for the Lagavulin and Port :)

Lalala, tommorow is my 0x18th birthday (24 for those that don't speak hex), for anyone in the Norwich area, I should be somewhere in the region of the Fat Cat pub from ~ 8PM in order to have some nice beer, all welcome, as always :)

It appears that the tg3 driver has been entirely removed from the debian stock 2.6.11 kernel images, this would be fine, except that that means the easiest way to get a working driver in debian, now, is to use the bcm5700 non-free driver. ARGH. Gimme back my tg3 driver, dammit.

In other news, decided to play with sshfs, seems quite cool, got the fuse module loaded and mounting remote filesystems over ssh seems to work quite nicely :)

May compile a custom 2.6.11 later to get back the nice shiny tg3 module, which is much nicer than the non-free driver I've currently got loaded. sigh.

Well, I've been up since ~ 6.45 ish, and was in Telehouse from ~ 10.45am through till 4.40pm (ish). Yesterday, our server hosted by netkonect in Telehouse "magically" lost routing to the world, I repeatedly told them that they had a problem, that the machine was fine before they had the other outage, and that traffic was consistently being dropped at their gateway... They insisted on rebooting our server, claiming that that would probably fix it... when it rebooted, of course, the routes and interfaces where set up exactly as before. Still no network. At 9.30pm Friday night we gave up and I resigned my self to going to London Saturday morning, got to telehouse, got to the right room, discovered that I hadn't got a key to our rack, or the monitor and keyboard that I'd asked for. So, phone netkonect, get a "oh, they should have given you the key at reception", right, trundled back to reception, yes, got the key, rah, now have access to the rack, but still no sign of a monitor and keyboard, or the netkonect engineer that was supposed to visit me, or the mistral engineer that was supposed to visit me. OK, fine, out comes the laptop, crossover cable, and some fiddling at the back of the server. Rah, I have ssh to it. Right, set up the second ethernet card to have a private address just so that the laptop can sit on that port and talk to it. Plug back in the other network lead, hey presto, still nothing from our box to the outside world, all being dropped at this point by the switch (I couldn't ping between the box and it's RAC port, which are on the same switch). Right, phone netkonect, tell them this, wait for a response, get bored, it's now midday and I'm dying of thirst, of course, the telehouse coffee shop is shut sigh, so off I go in search of coffee, ended up in the Bar-Cafe of Travellodge, about 5 minutes wander from telehouse. Sat there for ~ 45 mins waiting for responses, go back, and yes, I can now ping between the machine and the RAC port, WOO! Next step, right, yes, packets are being dropped straight after that, fantastic, so, our default gateway is down... back on the phone, long conversation, hang up, wait 30 mins to hear anything back, finally get told that our default gateway should be a COMPLETELY different address to what it has always been set to, 12 IPs upwards from where it was previously, on an IP that (fortunately) we hadn't already used in our range. SIGH, not happy about this, confirm that if I set the gateway to that IP it does infact work, but switch it back straight after confirming because I want THEM to fix THEIR router mis-configuration, after 2 hours of backwards and forwards getting nowhere I eventually caved, but not before making damned sure that they knew exactly what I thought of this event, and that if it ever happens again I will be severely pissed off, and that I want it in writting that their network engineers will not randomly change our gateway without consulting us ever again. I have a signed fax in my bag from netkonect saying that in the event of needing to change our gateway for whatever reason they will have to contact us to get approval. (I would have thought that was a bloody standard requirement anyway).

Anyways - all dealt with, networking came back, left telehouse.... It wasn't till a little while later that John sent a message asking if I'd remembered to restart the chroot web server (oh bollocks), no, I hadn't, but given it was running nicely until netkonect rebooted our server, and given that I'd forgotten by the point I left telehouse, I gave rough destructions over the phone, and it was all back up shortly afterwards (rah! thanks john).

On a different note, it was good to see you piem, you mad french man you :)

Right, having made a fair few changes in the last couple of days, I'm quite happy that the new version of bpgallery addresses the bugs that I found in 0.9.3. It's available from http://www.sommitrealweird.co.uk/development/bpgallery.html which now includes a link to my gallerys that I've generated using it.

The next thing to work on is a better set of documentation for it, especially for the themes. Basically a theme is defined as a set of bash functions that print the various parts of the HTML file or stylesheet, and can include some variables. It's a very simple way of doing it, but it works quite well (or has done so far).

My small gallery script has just gone through another release, the new version supports rudimentary comments taken from the file comments.txt in the photos directory, themeing support is almost complete, next release will have the destructions for how to write those, I hope.

If you're interested, it's available from http://www.sommitrealweird.co.uk/development/bpgallery.html .

Doing a very quick test of the speed, purely on the addition of keys in onak, we get the following results:

with keyd and fs backend on 100 keys:

$ time onak -b add < ~brettp/gpg-keys/splitfile-0.pgp

real    0m0.351s
user    0m0.004s
sys     0m0.007s

same data with the db4 backend

$ time onak -b add < ~brettp/gpg-keys/splitfile-0.pgp

real    0m0.106s
user    0m0.008s
sys     0m0.008s

and finally with the file backend

$ time onak -b add < ~brettp/gpg-keys/splitfile-0.pgp

real    0m0.031s
user    0m0.004s
sys     0m0.003s

interestingly, the simpler fs backend for onak appears to take a significantly longer amount of time to process a large import of keys that the db4 backend, tommorow I'll throw them each a slightly smaller set of keys (like 100, instead of 100,000), and get timings out of each of the backends for adding keys.

Still, I'm quite happy with the backends just working, and it being incredibly easy to change the backend I'm using (OK, so I don't do any data migration between the backends, but that would be reasonably difficult to do anyways, some of the backends don't implement iterate_keys, which is a shame).

I'm thinking about adding a mysql backend to complement the pg backend, patching up the fs backend to support iterate keys, and possibly implementing a caching backend that in turn loads one of the other backends as a keystore. I'm really quite enjoying working on this piece of software, it's certainly a lot less mundane than the day job :)