I bought a Visoly Flash Advance Xtreme 128M Card for the GBA and it turned out to be defective! Uggghhhhh... I have sent it back to the store for repair or replacement.
I have also bought a Lexar Media JumpDrive 256MB Secure USB Drive off eBay and am waiting for its arrival. Let us hope this one does not turn out to be a dud!
Yay!
I upgraded my home PC with some RAM (512MB that should have been recognised as 784MB - see my previous diary entry) and a 750MHz P3 bought off eBay and I can see the dramatic improvement this upgrade has brought to my system compared to the 128MB RAM and a 450MHz P2 that I had on this PC.
My system now scores 1280 on 3DMark 2001 compared to 585 before the upgrade! More importantly, GCC 3.4 now does a clean bootstrap in ~2 hours instead ~3.5 hours earlier!
I now eagerly await my GeForce 3 graphics card, also bought off eBay, so that I can enjoy Half Life 2!
The PC Hardware scene is crazy - my home PC is around 3.5 years old and it is already very difficult to find upgrades for it, especially CPU and RAM. I thought I finally got a great deal on two PC133 256MB RAM modules off eBay, but to my utter dismay found out that my PC recognised only half the capacity of each module!
After much research on the web, I found out that not all PC133 256 MB RAM modules are the same as far as the Intel 440BX chipset, that lies at the heart of my ASUS P3B-F motherboard, is concerned.
The most irritating thing about this is that there is no easy way of telling whether a particular PC133 256 MB RAM module is suitable for the 440BX or not, even by looking at the normal module details given on the site of the vendors.
When I had bought my ASUS P3B-F motherboard, it was one of the best and most scalable motherboards I could afford and it has already been obsoleted and has become the primary bottleneck in my system. This is so sad.
My PC is still immensely useful to me and performs most of the tasks that I throw at it quite well.
Yikes! GCC 3.4 takes 3 hrs 35 mins to bootstrap C/C++/Java on my machine compared to 2 hrs 30 mins for GCC 3.3 - that's an increase of 40% in compilation times between adjacent releases! Not good.
That said, improving compilation times is already a high priority item identified by the GCC maintainers, so by the time 3.4 is released, we should hope to see much better times.
Two clean bootstraps and a libjava testsuite run - and more than 8 hours of my weekend were gone. Uggghhh! :-(
Oh Joy! Mainline snapshots of GCC have also become available now apart from the normal release branch snapshots. This means that people on slow and unreliable links like me can keep up with GCC development.
On the other hand, trying to wade through two large underdocumented code bases - one at work and the other at home (GCC) - is somewhat taxing and I invariably end up too tired to hack on GCJ on a weekday by the time I reach home. :-(
Michael Tiemann's account of how Cygnus was created and sustained itself makes for fascinating reading! IMHO this company, now merged with RedHat, singlehandedly contributed immensely to the GCC/binutils/GDB projects and really laid the foundation for them to become what they have - we in the Free Software world really owe them a lot. Kudos to these cool guys.
I managed to get package-private access checking partially working again in GCJ - as Tom tromey had predicted, this did turn up a bunch of illegal accesses in libgcj that were not being caught all this while! Some of them are trivial to overcome, some not so.
Things like foo.Bar.snafu( ) still manage to slip through GCJ though if foo.Bar is a class in a different package with package-private access and snafu( ) is a method with public access. Ditto for fields.
This made me try to go through the front-end code to figure out where to make the change for checking this and Tom was proved right once again - though the parser proper is simple to understand, the analysis stage went straight over my head and repeated attempts to comprehend it have proved futile.
That doesn't mean that I am going to give up yet, but it does mean that I am filled with immense respect for people who have been able to work their way through the front-end code and actually improve it!
It also tells me how mistaken I was about my programming capabilities. :-(
For some wonderful and happy reason, tromey, mjw, Anthony and Jeff have all suddenly become active on GCJ again, all at the same time, after being dormant for several weeks! For some weird reason, they had also all gone dormant at about the same time.
This weekend I finally found out that the intermittent and increasingly irritating crashes on my home PC while building GCC on Linux was actually due to a defective RAM module - after removing the offending module, everything was back to normal, even after extended stress testing.
I do not want to code for Windows any more!
As a developer, I have always enjoyed working on Linux and have hated working on Windows. In the recent past, I have been exposed to all sorts of Windows ugliness trying to make GCJ work on Windows - I did not enjoy it at all!
This just is not my cup of tea - I love working on Linux and that is what I shall do.
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!
