I put out another Glom version, 0.8.14, fixing up the inline related records portal. This should be familiar to Filemaker users. And I'm starting to make the layout design a little more usable, though it needs a full Glade-like UI.

I made some updated screenshots with explanations.

Hacking on Glom is fun, and the code is become gradually less hairy. I'm feeling very productive.

For the last week or so, I've been herding the GNOME 2.10 release notes together, with much help from Davyd Madeley. The wiki was a really good way to get the initial information. As before, we use docbook for the actual text, and that lets us use Danilo Segan's xml2po magic to translate them just like people would translate GNOME applications, without suffering chaos when the original changes.

I've never found a good solution for multi-language web sites, so xml2po is the best reason I've ever seen to use xml as the original format for website content.

Luis Villa is taking care of the press release today, so it can be translated too.

GNOME's translators are incredibly effective, and this let's the 2.10 release show that.

I thought I was immune, but the 'flu got me too, knocking me out over the weekend and a bit extra. I seem to be OK now though.

This morning I had the first stage of having a crown put on one of my teeth, after it broke a few weeks ago. That wasn't fun either.

Back in August 2004, The C/C++ User’s Journal editor asked me to do an introductory article on libxml++, but has not replied to my emails since, or paid me for the delivered work.

So now it is online here instead as the libxml++ manual. At least now I’ll be able to correct any errors.

I just got my Dell laptop working with my new external flat-screen monitor as multi-head, more or less. I enjoy so much moving windows gradually from one screen to the other.

However, the Nvidia graphic cards don't seem to support the regular “Xinerama” way of doing multi-head. Instead they have TwinView, which I think makes X-Windows think that it has one big screen instead of 2 screens. I had to set this up manually in xorg.conf. I tried using the multihead options in Fedora's system-config-display, but that made xorg crash when starting.

I think that on a true multi-head system, secondary windows would open on the same screen as the application, but I think mine open wherever the cursor is. However, the GNOME panel is clever enough not to be wider than one screen.

And XINE gets confused. It squashes movies vertically, and I don't know how to stop that.

Note: In GTK speak, I might mean “display” when I say “screen” above. I get confused.

I bought a new tower PC (Dell Dimension 5000, with Intel P4 530HT) mostly to take the strain off when doing distchecks. Installing Linux was a bit of a struggle.

I wanted to keep the pre-installed NTFS Windows XP partition, but I did not find any LiveCD that was capable of resizing it. Most did not even recognise the (SATA) drive. The stable Ubuntu LiveCD would not even boot, though the “milestone” unstable Ubuntu LiveCD did. Knoppix 3.7 booted in default mode, but not when using the linux26 option. I gave up and used Partition Magic.

The Debian Sarge preview release was doing well, but then failed to boot after installation. For some reason it installed a k7-smp kernel, which I think is for a multi-processor AMD system, instead of a non-smp Intel x86 one. Stable Ubuntu did better, but hung while synchronizing the system clock. After googling, I reinstalled with the acpi=off option, which fixed everything. I wanted Debian on at least one PC, but Ubuntu won again.

I've been surprisingly successful with the gtkmm porting work that I'm doing for The Written Word. gtkmm,
glibmm, and libsigc++ now build with the SUN Forte CC, AIX xlC, IRIX MipsPro, and Tru64 C++ compilers, as well as g++ and the Intel
compiler. libsigc++ is quite a template test case, so that's a major improvement compared to a couple of years ago. I added a few configure
compiler tests to libsigc++ and glibmm to work around the (rather obscure) differences.

And a Microsoft employee even seems to have fixed a MSVC++ bug that
was causing us problems, though we'll have to wait for the new Visual Studio version to see whether it's really what we need. Back when I was developing on Windows there was no way to even report a bug, let alone expect it to be fixed.

After Hubert Figuiere mentioned it a while ago on Planet Gnome, I found a cheap SMC2835W 802.11g card on ebay, and am relieved to find that it works. It's a lot better than my 802.11b ELSA Vianect MC-11.

The only trick is to copy-and-rename the /Driver/smc2835w.arm file from the supplied CD, to /usr/lib/firmware/isl3890 (on debian linux) or /lib/firmware/isl3890 (That l is an L, not a one) on fedora linux, and then use the Gnome or Fedora networking control panel to create a new wireless connection. Maybe that firmware file is online, but how the hell would you know which file to use? I could not find any way to identify the “version” (1, 2, or 3) of the card.

I really hope that some dbus/hal-type thing could prompt me for the firmware in future. Looking in dmesg, googling for some dubious forum messages, and messing about in the terminal, is not a great user experience.

I'm very pleased that this works, but so far it still seems that no 802.11g card that you can currently buy as new will work in Linux, without recompiling your kernel or using the windows driver via the ndiswrapper hack.

Ubuntu just works

I also finally got around to replacing my debian partition with Ubuntu Linux. I was amazed that the 802.11g card just worked without any setup whatsoever. They seem to ship the firmware for many cards, and I hope they can keep doing that.

However, I can't get my new printer (HP Deskjet 3650) to work in Ubuntu yet, and that worked (almost “just work”ed) in Fedora Core 3, prompting me when I plugged in the USB cable.

Ubuntu is worth the install just for the little drum beat that greets you at the login.

The disabled root / sudo thing feels a little odd, and I think they should make it easier to find out what's going on. I mean, if I do su then it shouldn't just ask me for a password when there's no possible correct answer. As far as I can tell, on Ubuntu the regular user has rights to do some root-like things, but sometimes the user is asked for his own (not the root) password again. And, because su does not work, you need to run commands as “sudo thecommand” instead of doing su and then the command. sudo will ask for your own password again. It doesn't feel quite right, but maybe I'll get used to it, and maybe it'll improve.