Older blog entries for braden (starting at number 81)

15 Dec 2005 (updated 5 Jan 2006 at 07:07 UTC) »

I have a Sony G420 monitor that I purchased not too long after moving the the D.C. area. This is possibly the finest 19″ CRT ever made. I had problems with an NEC FP950 being too bright; the G420 is truly easy on the eyes. I guess I’ve been using it for more than 4 years now; and it still delivers a razor-sharp picture.

A coworker recently upgraded to a LCD flat panel and was looking to unload a Sony G410. I took it off his hands, figuring that it would be pretty similar to the G420 and thus make a good second monitor. My NVIDIA Quadro FX 540 can support dual CRTs, so I’d just have to plug it in.

And it does a pretty good job. The two monitors just barely fit side-by-side in front of me on my Anthro corner cart. Except for somewhat different sets of controls and ports, the housings are identical. So getting the images at the same height is pretty easy. The picture quality on the G410 is good; but it isn’t quite up to the G420. The colors on the G410 are significantly different and have taken quite a bit of twiddling to get reasonaby close to the G410’s “default” 9300 K setting.

All things considered, I’m pretty happy with the set up. Having the second monitor really is nice. But it also means that there’s absolutely no way I’m upgrading to LCD until nice roomy 1920×1200 displays are reasonably affordable.


So, it’s been more than a month since I set up WordPress and made my first few token entries.

Between work, the various obligations of the holiday season, and a raft of patches for OpenVRML, I’ve not had time for much. But there’s really a much lamer reason: I didn’t want to bother posting much more until I’d gotten Andy Wingo’s Advogato plug-in set up. Well, after much procrastination and fumbling about attaining the most rudimentary understanding of XMLRPC, I seem to have it working.

Last week I got e-mail from an old friend from high school who I’d lost touch with. Apparently he came across my comments on the NetBeans 5.0 beta. Of all the things. I probably wouldn’t even have written that if I hadn’t set up WordPress and been looking for an excuse to use it.


One, two…three?



Progress on the trunk has slowed down. There's some mysterious issue with transferring data to the plug-in subprocess that I haven't nailed down yet. And most everything else is suspended while I correct an unfortunate omission in the facilities provided by the API.

There will probably be a 0.15.9 release to address a few more 64-bit cleanliness issues that gcc 4.0 now trips over. I know this because …


… I am the proud owner of a new dual Opteron machine:

  • Tyan K8WE motherboard
  • Two Opteron 242 CPUs
  • 2 GB PC3200 memory
  • PNY Nvidia Quadro FX 540 video card
  • 3Ware 9500S-4LP SATA RAID card
  • Four Hitachi 160 GB SATA hard discs

I went with low-end Opterons as I was unsure of the price/performance payoff further up the line. And my intent is to upgrade these to dual core Opterons once the price on those drops quite a bit. As an upgrade from my dual 1 GHz Pentium 3 machine, the 242s seem quite snappy.

I've configured the hard discs as a single RAID 10 array. It's been a long time since I suffered a catastrophic hard disc failure; which probably means I'm due for one. I've wanted to do RAID mirroring for a while now; SATA finally makes it cheap insurance.

Installation of Fedora Core 4 went flawlessly. The RAID controller Just Worked; I didn't have to go fishing for any driver discs. Reading that 3Ware drivers were already in the Linux kernel is what prompted me to get the 3Ware controller instead of a Broadcom RAIDcore board.

x86-64 Linux on the desktop

Life in the AMD64-lane still has a few potholes.

  • Epiphany, my preferred Web browser, crashes often. It was pretty solid using i386 Fedora. Looks like I'm switching to Firefox.
  • No Flash. This one cuts both ways, though: at least 95% of the time, Flash animations are an unwelcome annoyance. But in a handful of cases it can be useful to have around.
  • The Sun JDK for AMD64 has no browser plug-in in it. Again, there's the occasional applet I'd like to be able to load. Fortunately, the Blackdown distribution does have the plug-in included and it works quite adequately.

I'm sure I'll find more such annoyances. But as that's all I have to complain about so far, I'm pretty happy.


At long (long) last, OpenVRML now has a nice C++ iostreams-based I/O abstraction. The library is out of the business of providing an actual networking layer; library users simply drop one in via the framework. The Mozilla plug-in uses this to load resources using the NPAPI, which takes advantage of the browser cache.

So… Long time, no entry.


OpenVRML now has a Mozilla plug-in. After a few incremental releases in the 0.15 series, it's finally more-or-less usable. And it runs out-of-process; so in the event it goes south, it won't take down the browser.

After more than one false start, the plug-in finally gives me a platform for improving OpenVRML's I/O story. More to come on this.

apenwarr: CORBA isn't about pushing around "objects" over the wire. (It can do that via valuetypes; but valuetypes are an ad hoc feature added mostly for parity with Java.) In general, CORBA "objects" have pretty much the characteristics of the interfaces you praise.

CORBA has some real problems. It's rather annoying to see someone who's just discovered IDL refer to imaginary ones as if they're well understood defects.


The good: as mentioned previously, the GNOME desktop is just more responsive; the autotools bustage I mentioned previously was not real (I just had some crufty symlinks in my project).

The bad: sound doesn't work (I have a friggin' Sound Blaster); Boost libs are missing the .so symlinks; the most recent kernel build gives me unusably slow mouse movement in X.

I wouldn't complain but that I get the idea that Fedora Core 2 development has been more or less frozen in this state for the impending release. Oh, well…Hopefully updates to fix this stuff will follow on the heels of the release.

Mozilla plug-in

It is now scriptable. Apparently plug-ins must now implement the NPN_* functions that were previously implemented by the browser in Netscape 4.x and previous. Now the plug-in must implement them to delegate to function pointers in a table gotten from the browser. This is implemented in the Plug-in SDK; but I'm not using the Plug-in SDK because it's a bit of a mess. It doesn't even appear to have proper releases.


I installed Fedora Core 2 Test 3. Nautilus is impressively fast; and I think I like the spatial mode. Surprisingly, 3D performance for my trusty Matrox G450 has also improved significantly.

On the downside, there seems to be some bustage somewhere in the autotools chain that's causing autoreconf to fail for me.

Mozilla plug-in

I decided to go ahead and do the legwork to make the OpenVRML Mozilla plug-in scriptable. Alas, my efforts have not been rewarded. For some reason, it appears Mozilla isn't making the calls to NPP_GetValue that it's expected to make in order to check if a plug-in is scriptable.

Mozilla plug-in

Success, finally, with the OpenVRML Mozilla plug-in.

The unsettling GLib assertion failure when destroying the GtkPlug was unrelated to the crash I was seeing. Instead, there was a shutdown problem in my GTK+ widget that just hadn't previously been evident.

Now to get the OpenVRML release out the door so I can check in the plug-in code.

72 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!