I just finished converting the GDK wrappers in gtkmm-1.3 to auto-generated GObject-based classes, which has
simplified the gdkmm memory model a lot.
We’re using a smart pointer that refs the GObject
automatically, and I was pleased to find that, with a
templated copy-constructor, it’s possible to use the smart
pointers polymorphically just like the underlying C++
classes.
This is consuming a lot of time. I need to get it done
and start looking for some paying work.
I’ve been doing a lot of work on gtkmm 1.3, and I just
mailed a request for some people to help out with some of the
easier stuff.
I’m becoming very optimistic about gtkmm 1.3,
particularly as I’ve succeeded in making the
wrapper-generation code
much easier to understand. In Karl’s absence I might be
going out on a limb a bit with some of my changes, but it’s
all in the name of making things simpler.
The job market is incredibly quiet. It’s going to be
difficult to stay in Munich if it continues like this.
I’m still making gradual progress on Gtk– 1.3 while I’m
searching for a new contract. I think that it’s time to try
building a simple example.
Oktoberfest started up again, with the usual madness.
It’s time to look for another contract. The project was cancelled after most of the permanent staff left. Now
there’s a lot of great reusable code that’s never even going
to be looked at. I’ll try to stay in Munich because it makes
no sense to leave just before the ski season. It’s not the
best time to be looking for a new contract, but I’m
optimistic that things are going to get better fairly soon
when everything settles down.
I have continued to clean up the gtkmm-1.3 code-generation perl code. It now uses
jamesh‘s more up-to-date and standardised
gtk.defs interface-definition files, so that’s another large
problem out of the way. I am now much more optimistic that
we can gradually put all necessary pieces together.
I turned down a Windows project at the company, for
obvious reasons. When I asked the project manager why they
couldn’t use Unix instead he explained that then they’d have
to give the source code away. Irredeemably inept.
Support the individuals who have lost loved ones. Don’t follow the crowd. Don’t start believing that newcasters and politicians are wise. Avoid blind faith in anything.
Work on my latest contract has entered a phase that I like to call ‘Whistling past the graveyard’. Meeting with management is like trying to see Catch-22’s Major Major.
Apparently Evil Robot Murray attended Guadec 2 on my behalf. Lefthand
side, in the blue, with the eyes.
I made a little more progress on Gtk– 1.3 today. I
figured out quite a bit about how the new code generator
works, added comments that should make it easier for the
next person, and started to implement the new signal system.
This <a
href=”http://www.linuxworld.com/linuxworld/lw-2000-04/lw-04-m4.html”>link
was very helpful when I was looking at the m4 code.
Flat-rate ADSL access is really making life easier.
Development requires frequent internet access.
I tried the Evolution snapshot and I am very
impressed. Outlook or the Netscape mail client are so
frustrating in comparison. Creating new filters for mailing
lists is so easy.
1 year later, I’ve finally got flat-rate ASDL from Deutsche Telekom – I am pleasantly surprised. And, unlike in the UK, it’s not
just for Windows.
At work, Roguewave have finally admitted that it wasn’t
fair to tell us to patch our compiler to fix one problem
with
their product, and then to refuse to support their product
with that compiler patch when we find another problem. Not
that they’re in much of a hurry to fix
it, though.
In the month that I’ve been waiting for Roguewave to get
their product working, I’ve
developed a damned good Sybase CT-library C++ wrapper to use
instead of DBTools.h++. I wish that I could release it as
open source, but I know that the company will never do that
even though it makes economic sense.