follow me on Twitter

    Blog has moved to, see January, 2005 on new site

    Blog entries for January, 2005

    This post moved to

    From Planet Classpath, Andrew Overholt jumps on the flash demo bandwagon showing 100% open source Eclipse from Fedora Core 4 Rawhide.

    This post moved to

    Dan Williams posted a comprehensive (and prioritized) summary of kernel fixes needed to make desktop wireless support work well.

    This post moved to

    I'm sure everyone else already talked about this but I was just watching the Daily Show from yesterday, and they had a screenshot of Gnomine as a news graphic. Apparently from ancient Red Hat, the one with the Red Hat logo in the window titlebars, not sure if that was in 7.x or all the way back to 6.2. Finally, gnome-games on TV! We have arrived.

    This post moved to

    Elijah, I think you may be even more screwed than you think. I recently went through some pain because I assumed that "a == b" would do a bitwise comparison of two doubles a and b. It turns out that this isn't true, and I bet that even if the compiler doesn't drop the i from "i+a>i" it's still not a meaningful test if you're looking for bitwise equality.

    Depending on the goal, you may have to write something like:

    double d1 = i + a;
    double d2 = i;
    if ((*(uint64_t*)(unsigned char*)&d1) != (*(uint64_t*)(unsigned char*)&d2))
      printf ("doubles are not bitwise equal\n");

    I was told the reason "d1 != d2" doesn't work for bitwise compare is that the x86 floating point registers are 80 bit so if one double is in a register and one in memory as 64-bit, then you get weirdness. Don't know if this matters for your application, but I found it surprising. (And it does matter for D-BUS since I wanted to check whether the right bits were marshaled/demarshaled.)

    This post moved to

    Miguel, I'm not sure interested is the right word in the context of that 94-page document. ;-) For years here is my mental model of floating point:

    1. To test numeric equality don't use the == operator
    2. If I ever write any numeric code, learn a lot more

    So recently I changed Havoc's Floating Point Fact 1 to:

    1. Never use the == operator with floating point, it doesn't even work bitwise


    This post moved to

    We're looking for operating system developers near the Red Hat office in Brno, Czech, including some to be part of the desktop development team. If you're interested, send an introduction and resume to Jay Turner (jkt redhat com). You must live in Czech for these jobs, please don't apply otherwise.

    This post moved to

    Luis linked to this Joel article which says:

    No matter how debunked Whorf is, I'm still convinced that Israelis are more likely to do things דווקא, simply because they have a word for it.

    Lots of people say Whorf was debunked because they read The Language Instinct by Steven Pinker. A professor of mine in college pointed out that while Pinker debunks a strawman hypothesis that language determines thought, Whorf never said that.

    At the start of his essay, "The Relation of Habitual Thought and Behavior to Language," Whorf quotes Sapir as saying:

    We see and hear and experience very largely as we do because the language habits of our community predispose certain choices of interpretation.

    Note the word habitual in the title of Whorf's essay, and the word predispose in the Sapir quote. The rest of the essay is consistent with those words.

    Read the chapter on "Mentalese" in the Pinker book where he goes on a big rant about Whorf. It's an attack on "linguistic determinism." He quotes this passage:

    We dissect nature along lines laid down by our native languages. The categories and types that we isolate from the world of phenomena we do not find there because they stare every observer in the face; on the contrary, the world is presented in a kaleidoscopic flux of impressions which has to be organized by our minds — and this means largely by the linguistic systems in our minds. We cut nature up, organize it into concepts, and ascribe significances as we do, largely because we are parties to an agreement to organize it in this way — an agreement that holds throughout our speech community and is codified in the patterns of our language. The agreement is, of course, an implicit and unstated one, BUT ITS TERMS ARE ABSOLUTELY OBLIGATORY; we cannot talk at all except by subscribing to the organization and classification of data which the agreement decrees.

    This is from the Whorf essay "Science and Linguistics" which is about the process of reaching scientific agreement through discussion. A conversation has to use language, and some things are easier and more natural to express than others. That's why jargon exists. As new concepts are invented, whether in science or business, people invent words suitable to discuss the concepts. Sometimes they skip words entirely and use pictures, mathematics, or even computer code.

    To me the Pinker way of thinking is sort of like what Alan Cooper calls "Homo Logicus"; someone who ignores the difference between what is possible and what is likely.

    The Wikipedia entry on Sapir-Whorf is interesting.

    I recently read the book On Intelligence, essentially a description of how the brain works. I'm not qualified to say whether this is an accurate book, but if so it would certainly debunk Pinker's "mentalese" claim that everyone thinks in a universal language of thought. Pinker argues that the brain works like the Common Language Runtime, compiling your language to intermediate language. On Intelligence sounds a lot more plausible than that, at least.

    This post moved to

    Christian, use APNG. MNG is seriously flawed. (We tried to do GdkPixbuf support for it back in the day.)