Older blog entries for djcb (starting at number 124)


I read wingo's lamenting against bonobo. Around 2001, Bonobo was quite trendy, and I spent quite some time playing with it, I even wrote some tutorials etc. (translated in to korean!). I read the Henning-book, I read the Don Box book, I read the OMG specs, and in the end, CORBA is not *that* hard (although it _is_ complex). Well, Bonobo makes CORBA a lot easier by hiding a lot of boilerplate, but not satisfied with that, I wrote some elisp to make creating Bonobo-objects ReallyEasy(tm).

Of course, if some technology requires two layers of sugar coating, it *is* complex. And you have to wonder if the complexity was really necessary, if it was really worthwhile. Well... looking back, I *do* think there is too much complexity in the system. Too many things can/will go wrong.

So scrap it all and replace it by D-BUS? Not so easy... For example, can we use D-BUS for communicating with DSO-objects? How to implement all the Stream/Storage stuff? How to replace the whole activation infrastructure?

Replacing something like Bonobo requires a lot of care. Yes - the system is overly complex, but a replacing system could quickly grow just as complex. D-BUS could make some of the things Bonobo does a *lot* easier; but AFAICS (and I am no Bonobo expert and a lot less D-BUS expert), some of the other things would be quite hard to realize with D-BUS, without creating a new monster. (E.g. D-BUS as the new IIOP, with the same complexity on top, but a lot slower...)


Thankfully, not all things in the world are so hard, and I released a new version of my little Python hack for view teletekst-pages; check my Dutch-language blog for details and pretty pictures, or go straight to the code (requires Python 2.3 and pyGtk).


Finally, I wrote a little program which makes sending XMPP-messages (Jabber) very similar to using mail(1) for sending, well, mail. Now, the servers I maintain send me IM when they need my help... E-mail is soooo 20th century...

long time no advogato. I am still staying in the Netherlands (The Hague). More-or-less finished a project doing crypto stuff, and now doing some source code archeology :)

Recently, I have updated my gnuvd Dutch dictionary tool. I was happy to see Debian to update their version as well with only three days! Also, some googling revealed some users in various weird places, like an S&M forum :-)

During medieval times in Gnome history (1999), I wrote TeleGNOME, to watch Dutch 'Teletekst' (videotext) pages by leaching them from some website. Now, I re-implemented the program, using pygtk, which has been a delight so far. Glade leaves something to be desired, of course. But it does the job, and it's definitely better than writing the XML by hand :-)

Here is the code (tb.py), which has everything you need. Requires Python 2.3 I think.

If you can read Dutch, there is more info on both programs in my blog.

Maybe my next project should be something with Mono; it's an interesting project, and all political considerations aside, I can't hurt to hack a little thingie in C#.

I am wondering how much work it would be to connect mutt to the the whole Evolution Data Server (EDS) infrastructure; I have that I can process mail in mutt faster than with any GUI mailer, but the EDS-stuff is pretty cool, and it would be *great* to have all my apps share the same addresslist etc.

back. Back in Holland after nine months. Hmmmm... Australia, Malaysia, Thailand were better... But at least I can do some proper hacking while I am back home. I released a new version of webtex (nee webmath). It enables you to use LaTeX on the web, in particular you can put the beautiful LaTeX-formulae in your webpages, like this:
<img src="pyth.png" name="a^2 + b^2 = c^2">
And webtex will create pyth.png for you. But no need for non-standard or non-supported (such as mathml) tags.

Now, I have added support for the beautiful Thai script. Check the webpage for more details.

kai: a good start may be the elisp I wrote to generate the boilerplate for Bonobo and Gtk-widgets; check here. Note that it's all GTK+ 1.x stuff, and especially the GTK+ stuff is not well-tested. But anway, it may be helpful: gwizard
webmath: mathematical formulae in webpages I was a bit frustrated with the difficulties of displaying mathematical formulae in webpages. So, I wrote webmath, a little script to bring LaTeX/TeX formulae to the web. It has the following features:
  • use standard html; no browser extensions needed;
  • also, no magic tags for some preprocessor, your html page will not be rewritten
  • use TeX/LaTeX syntax (insert MathML example:-) and png images
The trick is to do something smart with the img-tag: <img src="abc.png" class="webmath" name="a^2 + b^2 = c^2">

Now, webmath will generate abc.png for you, with Pythagoras little gem. It works really easy, and I have an emacs minor-mode to make it even easier... Enjoy.

thoughts. I did not really know either Ettore or Chema, but met both of them at GUADEC in Paris. Nothing can compensate their loss, but they will live on in the things that they created.

Played with the Perl/GTK2 bindings a bit. They look very sweet; they may not have reached the quality of, say, the Python bindings yet, but they will get there. I appreciate the work murrayc is doing with the GNOME-Bindings. Scripting languages + GNOME-bindings are great for RAD, and making the bindings easily installable, universably available will allow for more software quickly. I still haven't figured out how to implement drag&drop using Perl/GTK2. The documentation is a bit lacking in this respect. I am writing a little program which will accept URLs (for example as seen in emails) I throw in there (using d&d). It stores them, and when I am online I can click a button to open 'm all in my webbrowser. This is one of the tools I am writing to make Linux life a bit easier when you are only occasionaly online (travelling laptop users such as myself...)

Another program I wrote, also for the occasional online usage, is a simple script on top of httrack, the program which mirrors websites so you can read them offline. Well, it's a bit inflexible in that you cannot (easily) download multiple sites with different settings for the depth of the internal and external links that should be downloaded... so my programs does that; just fill in the websites and the desired depths in a config-file, and the program does the rest for you... some what. Needs more testing and maybe a GUI.

Well, enough :-) Wetter's disappointing here in Melbourne today.

mutterings. After experimenting with various way to handle the daily influx of e-mail, for the time I am staying with Mutt. I filter all my mail in either (1) bulk e-mail, (2) private e-mail and (3) spam, using some procmail[1]/spamassassin magic to sort things out. Only my private e-mail arrives in my 'inbox' email folder, and won't leave that folder before I handled it (replied or decided to ignore :-). In mutt I have some simple keybindings to move handled e-mail to either 'business-archive' or 'private-archive'. Thus, I never forget to reply to messages, because there are never more than 10-20 email in my 'inbox'-folder.

Because I can do *everything* with keybindings, and define my own as well, mutt is really the fastest way. One of the weak points of mutt is that it's not possible to see other e-mails while you are writing one - but I just open another instance. I use DJB's Maildirs (instead of mbox), which stores e-mails in separate files, so concurrent mutts won't corrupt my e-mail. Another problem with mutt used to be it's slowness when opening big e-mail folders (like 2000+ e-mails); however, the header-caching in recent mutts has solved this.

Mutt allows for a lot of customization - but if it's still not enough, there are now guile-enabled mutts available...

I have talked about travelling with my laptop before. I think I should write down the things I have learnt about it - the Linux-Travellers-Howto.

Footnotes: [1]: don't read procmail source code while eating.

despair and hope.Finished my little mod_rewrite-job but it took TooLong(tm). The mod_rewrite syntax... leaves something to be desired... and external mapfiles must return "NULL\n" upon failure, not "NULL" as the docs seem to suggest... oh well. However, the reason I could get the job done at all was using Emacs and tramp-mode... I'm working here from a Sydney internet-cafe on a computer located in Amsterdam, and for some reason, it takes about a second for every keystroke to go from Sydney->Amsterdam andd then back from Amsterdam->Sydney (using vi in my terminal). And as I am a lousy typist that is quite frustrating. But 'tramp-mode' makes a local copy of files I edit, and copies them back to Amsterdam when I save them. ssh foo@bar 'command', especially since I use the identity.pub/authorized_keys-trick, so I don't have to type the crypto-friendly-means-human-hostile passwords all the time.

What are the options for free-software dictionaries? My debian offers a lot of dict-packages, but so far I have only found the dict-gcide-package of good quality, and that is mosly the 1913 Webster's dictionary!

I am reworking all my webpages... it's both interesting and boring. I have a bunch of pages, and I'd like to preprocess them offline, as my www-provider does not allow dynamic webpages... The last time, I used Latte for preprocessing my webpages; well, Latte and it's successor Blatte seem rather dead. So this time I want to use something else. I'll probably use xslt-proc. XSL seems like the fashionable way to do it 2003/2004.

I found Philip Lord's pabbrev-mode for Emacs, which dynamically suggests expansions. It will gather suggestions from the current buffer. When I type "inte" it shows "inte[resting]", and a TAB confirms. Wow - this saves me typing!

Still doing some remote consulting for a dot-com back in Holland. Taught me a lot about Apache, MySQL and PHP... (and I would choose PostgreSQL and perl if I could choose... anyway, still some mod_rewrite magic to do...

As I am in Sydney, I should meet up with the SLUG... hmmmm.

My Debian-unstable seems painfully stable these days... nothing to apt-get upgrade for more than a week. Must be because of the hack... I would be very interested to know why anyone would want to hack into Debian or the Linux kernel - it seems contrary to hacker ethics, and there is no monetary gain tot be made... unless your money comes from proprietary competitors, I guess... <insert conspiracy theory>

reading. I am one of those sorry souls who brings his laptop computer, even while backpacking. Actually, it is quite handy to have all my stuff (photos, music, email) near me. And nowadays, I can usually plug my machine in into the local internet-cafe's network and exchange data with the outside world. Nevertheless, I spend quite some time off-line. I will talk another time about what software tools make this nomadic computing a bit easier, but want to focus on reading material now.

I love reading. Both fiction and loads of non-fiction. But of course I don't want to carry 20 kg of books with me. Below are some of the electronic books I found. All are free for download (or browsing, but httrack and friends help...), but the licenses differ wildly. And some are PDF, some are HTML or txt. It takes some time to get used to reading books from a computer screen, but it's worth the effort!

  • Dive into Python. Python for programmers. Nice refresher course. It's even apt-getable.
  • Handbook of applied cryptography. Skimming through it, it seems pretty solid. But I haven't really read this one yet.
  • Introduction to methods of applied mathematics. Sean Mauch's excellent book on mathematics. Far from being a mathematician, I found his book a good way to freshen up / expand my math skills. Well, I only read a part, its 2300+ pages!
  • Simple nature. Benjamin Crowell's introductory/lightweight book on physics.
  • The Art of Unix Programming. Eric Raymond's (ESR) attempt at describing (and prescribing) the Unix programming philosophy. Even if I don't always agree with the book or with ESR, ESR is a good writer and the book is an interesting read.
  • Free as in Freedom is Sam Williams' biography of Richard Stallman. A fascinating book.
  • Nikola Tesla's Autobiography. About a man I would call an 'applied genius'; he never really got the recognition he deserved, and man was hestrange.
  • Hacker crackdown - Bruce Sterling's account of the 1980's hacker (i.e. cracker) scene. Great read!

Now, what I am still looking for are a good general work on chemistry, on statistics and on discrete mathematics.

I'd like to compile a longer list, preferrably with only 'free'/'creative commons' - licensed books. If you have suggestions for this list, please email me.

115 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!