Older blog entries for jonas (starting at number 26)

Birthday tomorrow so doing laundry and stuff today instead. Noticed in the process that I need to find a dentist soon because my wisdom teeths are beginning to become a nuisance and I'm getting crankier and crankier the more my body deteriorates. I think I'm ready for my cane soon :-)

Thought of the day: email. What thomasvs wrote made me think about a discussion I had a few days ago. The topic of the discussion was email and, in particular, how clueless users are trying to make it do something it was never intended to do and administrators are trying to make them understand why it was never intended to do so.

It struck me then that noone, to my knowledge, had decided to implement email using XML. This seemed odd to me since everyone and their grandparents seems to want to do everything with XML these days. So I had to look it up. It turns out that Graham Klyne wrote a draft for the XML emails in January. I'm not the least bit surprised at this.

The second topic of discussion was the large attachements that users like to send and that's causing all sorts of grief. I think that we have to accept that users will keep sending messages as attachements, no matter what we think of it. The reason is that it's a fairly natural process to do so. The problem is not the emails themselves it seems, but rather the framework that has been built around the email services. The nature of an MTA is to be dumb. It will accept a message, forward it somewhere or deliver it and then forget it ever saw it.

Now, imagine for a change that the MUAs and MTAs were a bit smarter. Imagine that MUAs had a reliable way to tell the MTAs that "to this message, the user wants to attach this, this and this file." The MTA could say, "send them to me" or "no, I already have those". When forwarding to an MTA, the exchange could be something like this:

  Sender: Here is message A. Message A has attachement B.

  Receiver: Okay, I've got message A. Please send attachement B.

  Sender: Here is attachement B to message A.

  Receiver: Okay, I've got attachement B to message A.

  Sender: Here is message C. Message C has attachement B.

  Receiver: I already have attachement B.

  Sender: Here is attachement B to message A.

  Receiver: I told you I already have attachement B.

            For your crimes you shall not be allowed to eat tonight, nor shall

            you be allowed to send any more attachements for three days and you

            must remain in your address space where you will contemplate the

            errors of your ways until the full moon rises again. Now begone!

Okay, so it's not a very likely scenario, but still. It's not the users fault that they want to use the features of their software. It's the industry that has adopted features without understanding their implications on the underlying framework. I hope this changes in our lifetime.

The job I went to look at looked really, really fun. I'll hear from them again first week in december so we'll see then what happens. When I got back I spent some time oggifying some CDs and played musical chairs with some larger files to make them fit in their proper places.

Not a lot of things happening otherwise, except an email from Lockheed-Martin telling about their use of the GNU Libg++ in an Air Traffic Control system for the FAA and requesting input on future availability of the product. That was interesting.

So then it turned monday again. Today has been relatively unproductive since I still havn't gotten into any high gears. Hopefully, that will not be a problem tomorrow since I have a job interview quite early.

What I've been toying with today is various DNS visualizations. My first feeble attempts made me realize that I'm about 1GB RAM and 1GHz of CPU too short, but I think I've managed to overcome that now and the visualization software only eats about 547 MB RAM. Of course, that's for a limited map of about 900 nodes and corresponding edges.

The software outputs its information as Postscript files, which is generally nice except when they contain 900 nodes and corresponding edges. Ghostscript failed miserably and Gimp took two hours and then crashed telling me that it was "unable to seek to tile location on disk". It took the rest of my desktop with it too.

After sorting out those problems I spent a few hours investigating some statistical algorithms and methods and realised just how little math I attended at the university. Had I spent about two years more there I would likely be able to whip up some much faster algorithms for the work I'm trying to do. Right now I'm substituting fast algorithms for plenty of CPU time.

Somewhat related, I got around to setting up a home directory for me on my AFS server, so I can now push things into AFS space. I also have a tool that basically just takes a command, pushes information about it over to another machine and executes the command when there is CPU time available and then mails me when it's done. This helps in theory those long tedious waits for things to finish and I can crash my desktop to my hearts delight without affecting those runs.

This was in theory. In practice, I do them on my desktop machine anyway since the only other available machine is a 486dx33 with a constant load of about 6 or 7 since it's also doing other tasks. Of course, it doesn't really help that the machine in question has 48 MB RAM in total, and the things I want to do need more than ten times that much.

In unrelated news, my girlfriend was delighted today as she received her first SirCam virus in Spanish. Personally, I've been spared most spam today which I think is a first for a long, long time. I didn't escape the spam from the postal office though and I had to sort through a pile of paper until I found a notice from my bank telling me about a deposit to my account. Now I need to get to the post office tomorrow to drop off another load of Penguin Mints.

So I got back two days ago, on the 16th, but hasn't really gotten into gear until today and it's still in a very low gear. The work in Brussels ended a day earlier than I anticipated, so I took an early flight back on friday morning. Yesterday was spent coming to terms with the email that had accumulated over the week and catching up on other issues, like sleeping.

Today, I've spent some hours investigating VAT regulations with trade of services inside of the EC and came to the conclusion that I didn't have to worry about it, so the invoice I sent for the week ended at roughly USD 2600, so that felt good anyway. I need to find ways to send more such invoices in the future if I'm to survive though.

If I can allow myself some self-pity, I have about USD 300 that should last until january. ;-) We'll see how that sorts itself out, but things has a tendancy to work out some way or another.

No earth-shaking events really took place during the past week it seems. My birthday is this upcoming thursday though, but no great plans for it except maybe dinner somewhere nice. So in either way; have a nice continued sunday everyone! :)

Most of yesterday and today was spent in a frenzy trying to get things ready for my departure tomorrow. Updated the Debian installation on my laptop, fooled around with Linux kernels 2.4.x and PCMCIA. Finally got things back together again and is now enjoying a computer with an up-to-date installation of most of everything on it. Re-configured Gnus (I seem to do this every time I go away..) and set it up to work in unplugged mode.

Copied all the sourcecode I'm working on to the laptops home directory, made fresh CVS checkouts, installed all Perl modules I could even think of a faint use for, just in case, and rejoiced when I got a message telling me that my PAUSE account is ready for use.

I'm leaving tomorrow and won't be back until next sunday, so don't expect much updates until then.

As predicted, I didn't get to play much with C++ today. I've decided on a pet project though; adding some functionality to an existing C++ project.

goingwhere: In response to paypalwarning.com, I'd like to say that they're probably right in everything they say. I don't really use Pay Pal, fortunately. Sometimes I still think it would be good to find a US bank account though.

Anyways, the problems that paypalwarning.com mentions about their support reminds me of a problem I've had with Juno Internet Services. You see, every now and then I get a message from their abuse department telling me about a spam message that they say originated from my network. Looking at the headers, the spam messages always come from completely different networks.

Not only is this a problem, but they also send the messages to bug-jwhois at gnu.org, which is an address specifically for reporting bugs about the GNU jwhois program. This had me confused for a while and I didn't receive a reply to my inquiries sent to them aside from an automatic response telling me they'd look into whatever it was I wrote to them about.

After a few months though, I figured out that they must be using jwhois in their backend systems for finding out who to send abuse reports to. When they call jwhois with invalid options, or no options, it will answer with some information telling the caller about the syntax he or she should use. This syntax ends with this;

Report bugs to bug-jwhois@gnu.org

Their system apparently matches this email in the belief that it's a valid address to report spam to and offers it to the abuse person who then sends the message to me through that address.

Now, I still havn't received a message from them about this, and I must have mailed both Juno and their Internet uplink 6 or 7 times over as many months. I havn't received a letter in almost a month though, so I'm hoping that the strongly worded letter I sent to them via the normal postal service actually made a difference. Of course, I still havn't received a reply to that message either.

Regarding the job I went to look at; the woman I spoke to said I most likely had the qualifications for the job but had problems with two things; (1) that I don't have a drivers license yet, and (2) that I'm apparently about six years too young for their taste. The second one I can't do very much about, but the first one I can and should. I only need to find about $5,000 or so to get my personal economy back on track. This should be fun..

This has bugged me for a while, but I havn't bothered to rant about it yet. Paypal. I have an account (with the imaginatory name jonas@gnu.org), but I have no way to withdraw money from it (yet). The reason? With Paypal, you can only withdraw money to a US bank. This means that I (a) have to find a US bank to open an account in, and (b) find a way to transfer funds from the US bank to my normal bank.

Why is this impractical? Because I'm in Europe. I can't very well fly over to the US just to open a US bank account to use for withdrawing Paypal funds. The solution? Find a bank that allows foreigners to open accounts from abroad. This seems pretty much impossible.

Next rant; C++. I decided that my first step will be C++, from which I will then evolve into learning Java. The reason is that I took one look at C++ and decided it is crap, so I might as well learn it first and then be able to appreciate the beauty (relative!) of Java, instead of being depressed by the uglyness of C++ after having learned Java. Maybe I'll find time to finish up C++ tomorrow, but there's a possiblity it will take until friday until I can find the time for it.

Tomorrow will be spent on two tasks; first going to a drivers school and getting some papers so I can eventually (within four years) get a drivers license. Eventually, I may need a drivers license if I'm to work more than 10-20 kilometers from home. This leads to the second task of tomorrow; visiting a city roughly 70 kilometers from where I live to see if the job as IT manager is something for me.

And finally, something to relate to yesterday; after I wrote the entry I spent some more time tinkering with OpenAFS and eventually got it to work with surprisingly little effort and only two or three reboots. Not only am I able to access my own AFS cell, but I've also made a mountpoint for Stacken's (stacken.kth.se) AFS cell, so I can see their public information, which is a nice nicety of AFS.

Someone that posted a diary entry here on Advogato was working on domain registry stuff. Who was that? Show yourself! I've lost your name and I'm interested in knowing what you're up to.

Also, I've decided to wet my hands with Java. Beware.

Having TAGS files for most of your sourcecode really improves your coding. There's only a few things I miss, but I may implement them tomorrow. Should be a nice exercise in elisp and will make it much easier for me when working on multiple sub-projects simultaneously. Command of the day is therefore M-., and if you havn't bound your Window keys to M yet, you should do so now. (the menu window thingy key is bound to Compose of course).

Also been toying with Coda and OpenAFS and looked at Arla and OpenGFS. I've pretty much ruled out any AFS implementation for my purposes, but I'll try to make OpenAFS work just for the heck of it. Then I might also toy with OpenGFS to see how that works in depth, but I'm telling you, compiling everything on a Pentium 133 is tiresome.

kevindumpscore: The standard you're looking for is called TeXInfo :-)

I'm writing this entry early, because I don't expect anything even mildly interesting will happen today. I have now received tickets in the mail though for the trip I'm going to take next week. It will keep me occupied for pretty much most of the week and I won't get back until next sunday.

In other interesting news, someone told me that I ought to apply for membership in the GNOME Foundation. I'm not sure if I would qualify or if I would have anything to do as a member though. Opinions welcome.

Looked at XML/XSLT/XHTML. I'm still not convinced it's a good idea. Or rather, I think it's a very good idea, but I'm not convinced it will catch on in the industry at large. I've still got a fresh mind about it though so we'll just have to wait and see. This relates by the way with what I said about that all content management systems suck. No system I've looked at for the web, like Zope, PageKit, AxKit or whatnot, has really interested me. I'm sure they're good, but their use seems limited to me right now.