After resigning from Zarlink (and rejecting Google) I'm still unemployed! This is despite looking for five months. The IT sector in the UK/Europe seems to have collapsed in on itself, and apart from finance/public sector infrastructure projects, all work seems to be in either Digital TV or mobile phones. Unfortunately after working in digital TV for three years I'm looking for something different. Seriously if anyone knows of Linux kernel work that can be done from the UK please get in touch!

I've done a lot of tidying up on Squashfs (open source compressed filesystem) and I have made some major improvements to directory handling. When I originally wrote Squashfs I envisaged it mainly being used in embedded systems which tend to have filesystems with small(ish) directories. Due to this I limited directories to a maximum size of 512K and I knew handling of directories over 64K was slow. As with many things the uses that some people have put Squashfs to have surprised me, and I've had complaints where people have tried to archive directories with 100,000 + entries with Squashfs. Needless to say the original design didn't work terribly well! I've completely over-hauled the directory design, and I happy to say the next release can handle directories of many megabytes in size with no performance loss.

It probably goes without saying that the last few months of unemployment have shown the importance of having an open source project to work on. Before I started my open source work after leaving a company I had literally nothing to show for my years of work at the company, and no programming of importance to do until I got my new job. At least now with Squashfs I've had something to work on during the months of unemployment!

I am now unemployed after resigning from Zarlink. I didn't intend to become unemployed because I had a job offer from Google! Everyone raves about how good Google are to work for, however, they expected myself to relocate from the UK to Mountain View California, and I was not prepared to do that. The whole experience with Google was a huge waste of time, I originally applied for a job in their Zurich Research labs (they make a big song and dance about how they're trying to attract Europeans not prepared to relocate to the States by opening these labs). I went through the whole drawn-out interview process including a trip to the States (and this was not a fun trip because I dislike the states, and their attitude towards foreign people). They offer myself the job but decide I have to go to the states. I resigned from Zarlink when I received the offer expecting myself to go through with the relocation, it was only after this I thought 'why the hell should I relocate, Google may be a good job but it's not that good'.

Squashfs my open source compressed filesystem for Linux has done remarkably well over the past year. In fact it appears to have become the de-facto compressed filesystem for all new embedded work, and lots of routers, ADSL modems, set-top boxes out there are running out of a Squashfs filesystem. In addition about half of all liveCDs now use Squashfs as the filesystem, and it is slowly replacing cloop, having already replaced zisofs and cramfs...

Notwithstanding all the "success" of Squashfs, I regularly wonder whether I'm going to do another release. Even though there are lots of companies using Squashfs, I have received a princely $20 in donations, and so it's a good job I didn't do it for the money. What payment I do get is in 'ego boosting' but even then a large majority of the users of Squashfs never acknowledge they're using it, or that a lot of what they've done would have been impossible without it. Of course the most irritating are the ungrateful users who download it (for nothing), and then email you to say "very nice, but what about this and that feature, why haven't you implemented that", strangely forgetting in this life you get what you paid for.

In short I'm disillusioned with open-source and the open-source model. I can't help thinking back to the 'old days' when people expected to pay for software. If I received money for Squashfs I wouldn't be in the position I'm now in, unemployed, no income and spending my savings in living expenses.

No new entry here for quite a while - work work after Christmas has been pretty hectic with all the managers promising to get things to various customers with silly deadlines. Nothing changes.

As a relaxation (nerd that I am), I have started on the code that will form the next release of Squashfs 1.2, which adds off-line append capability to the read-only filesystem. I always intended to add this, but a couple of people and organisations have already asked for it, which is good, as it means I know people will use the additions.

I have always been interested in finding out what is the smallest usable Linux machine for carrying about. To add to my toys, I have bought an Ipaq 3950, which potentially allows you to have Linux in your pocket... I have put familiar and Linux on it, and as expected, it works pretty well for a box to play with, but it is useless as a box for doing any Linux development. This is is because as a standalone system (as opposed to sessions running remotely over IP over PPP & USB) it is useless with only a touchscreen and stylus. The foldable micro keyboard I bought for it, would probably make a more usable system, but the 39xx being a newish machine is only just getting an alpha Linux kernel and drivers. There exists no support for microkeyboards currently, which means I'll probably write one. So, the jury is out, and my 12" iBook is currently the smallest non-toy Linux system I have.

Thanks to all the replies to my post about responding to unhelpful messages about your open source software. As I thought, most of it is common sense, but it is good to get a reality check, and confirm that your common sense reaction is in tune with other people.

I've been reading Salmoni's diary. You mention

I would have learned [Java] a few years ago because my local college was offering an evening course which I could get onto cheaply, but they insisted on using Microsoft Java (was it called J++?) which even then I knew wasn't the standard so I didn't take the course and went to learn Python instead. I wonder why they insisted on using MS's version?

The tendency (in the UK at least) of Universities/colleges to teach/instruct in Microsoft products is annoying. As to why, it's the joint problems of business pressure and budget cuts. When I first started out as a PhD student(1989), the computing department where I worked (no names to protect the guilty) used Sun workstations/multi-user servers to teach undergraduates. However, with the pressure to cut costs, we moved over to PCs and Linux in about 1995. So far so good, however, with the department now using mainstream equipment, it was argued that the department didn't need it's own equipment and could use central university pool (of machines) provision. The problem here is pool machines use Microsoft operating systems, and therefore teaching had to be done under Microsoft operating systems.

From here it is a slippery slope towards total Microsoft product use, with the encouragement that Microsoft of course sells products cheaply to educational establishments, and that students get a 'skill' in demand by business.

As a lecturer (assistant/associate professor) you are often powerless to resist such changes, the changes being forced on you from high, for purely financial reasons. Of course the people forcing the decisions are unaware of the differences between a general purpose machine and one for computer science teaching.

As far as the 'slippery slope' goes, it saddens me that my old computing department appears to now be pretty much a Microsoft shop, they are now getting funded by Microsoft for research, and recently an ex-colleague of mine went over to Microsoft HQ to accept some sort of academic cooperation award.

As an aside you (Salmoni) mention you used a Dragon 32, and that you now work at Cardiff University... Did you know that the Dragon was not a complete copy of the Tandy CoCo, and as such, the keyboard driver and other required modifications to MS Basic, was done in the Computer Science department at Cardiff. The machine was also manufactured not many miles from you in Port Talbot (opposite the Margam steel works).

The Dragon was the first computer I had. Like you I could not afford to buy an assembler, and I still have a notebook from 1983, with machine code laboriously worked out by hand. I finally wrote my own assembler, which was then the longest program I'd ever written!

Another birthday, another irritating day at work, trying to do the work of 3+ linux kernel developers with ever more unrealistic milestones. Look on the job sites and find there is exactly one linux kernel developer job on offer in Europe, on the other side of the European Union. That's still one more than there has been. Of course, there's lots of kernel developer jobs in America as usual, senior ones too requiring a PhD, rather than the crumbs on offer in Europe.

Slightly bemused by the fact I've still got no certification, even though I have been certified three times. Having read the blurb I can see how that might happen, but it seems rather silly.

No open source development over the weekend, just a lot of boozing. I will probably start on the next set of improvements to squashfs (1.2). It's good to get feedback, and I've already had people asking for the improvements I was intending to do.

Chalst asks for suggestions/feedback on his list of seminal highlevel languages. I must admit I disagree with Occam being a seminal language - the elegant parts of Occam were straight from Hoare's CSP (Communicating Squential Processes). The other parts of Occam were a nightmare, no dynamic thread/memory or array allocation, no structures, no shared memory between threads, and a intensely irritating syntax straight from hell (Pascal). It was in short a brain dead version of Pascal with Hoare's elegant CSP added. The real innovation was the implementation of CSP's channels in hardware on the transputer, a concept which made plugging together multiple transputers into a parallel computer network easy.