Older blog entries for haruspex (starting at number 118)

5 Oct 2004 (updated 6 Oct 2004 at 14:05 UTC) »
shlomif,
I recently had to figure out how to do something with wget: get everything below a directory on the web-server, without following links to outside it. So I logged in to FreeNode's #debian channel, where there are many knowledgable people to ask it. The first answer I got was "RTFM". ...
No matter how much you bleat, the answer has always been in man wget AND wget --help. (The latter is impossible to miss, as any incomplete or incorrect invocation suggests you use it.) If you need IRC to babysit you through a command that has comprehensive built-in help, well, you deserve what you get. Make an acronym of this: Try Harder Next Time.

(update) nymia:

So, what is shlomif's chance of getting the answer: 25%.
100% if he uses man wget or simply wget itself.

IRC was the wrong avenue, as he may or may not learn. The fault is not IRC per se. "RTFM" is a valid response because what is required here is not the answer asked for, but the knowledge of where in general to find such answers. "Give a man a fish" versus "Teach a man to fish". The former does not scale. The latter does (and is the premise of education in general).

To your cases I would add:

5. Whether the respondent knows the answer is irrelevant; "RTFM" is appropriate when it is clear from context that answering the question of the instant will do no lasting good.
4 Oct 2004 (updated 8 Oct 2004 at 16:17 UTC) »
raph -
I now have a rough interactive app for drawing with Cornu splines. ... I drew the letter R freehand,

Your editor looks very nice. And the "R" does too. Doubles as a good demo of libart - which I've been wanting to play with too. Well done.

30 Sep 2004 (updated 30 Sep 2004 at 04:20 UTC) »
apenwarr: have you taken into account:

1) that certain features (or lack of) in some compiled languages militate against reliability (and sometimes even performance)

2) that certain features of "scripting" or "interpreted" or "the other sort" of languages are actually designed to effectively aid reliability

I raise this not in order to settle the debate - which is clearly not a convergent debate as has been proven more times than there are seconds left in my lifespan - but in order to perhaps make you think for a moment. Your assertion is just a troll in a skimpy kimono.

27 Sep 2004 (updated 27 Sep 2004 at 18:18 UTC) »
bolsh wrote,
in some locales the alphabet is ordered "AaBbCc....Zz", and in others it uses the ASCII ordering of "ABC...Z...abc...z". This depends on the LC_COLLATE environment variable
Aha! I never would have thought of a non-ASCII collating sequence. So herzi's original problem might have been caused by a collation aAbB..zZ?
26 Sep 2004 (updated 26 Sep 2004 at 21:53 UTC) »
herzi: Curious sed problem. I am not a sed guru - although I always wanted to "get my head around it" - but I learned the following.
  • GNU sed is case sensitive by default. The man page for BSD sed is coy on case sensitivity and doesn't mention any switches for it. It appears to be case sensitive.
  • OS X has BSD sed of unknown version, so no GNU extensions like '+' in regexps
  • You need to use sed -r to get the extended regexps (see info sed section on extended regexps)

    Neither version showed the [list] problem you described:

    (Gentoo)
    toby@jumpy toby $ echo FOO|sed -n -e 's/[A-Z]/X/p'
    XOO
    toby@jumpy toby $ echo foo|sed -n -e 's/[A-Z]/X/p'
    toby@jumpy toby $ 
    
    

    (OS X) imac:~ toby$ echo FOO|sed -n -e 's/[A-Z]/X/p' XOO imac:~ toby$ echo foo|sed -n -e 's/[A-Z]/X/p' imac:~ toby$

    I was using GNU sed version 4.0.7 (sed -V) on a Gentoo box, and OS X 10.3.5.

    Other sed tricks I amused myself with:

    toby@jumpy toby $ date|sed -r -e 's/[0-9]/X/g'
    Sat Sep XX XX:XX:XX EST XXXX
    imac:~ toby$ date|sed -e 's/[^ ]/X/g'
    XXX XXX XX XXXXXXXX XXX XXXX
    imac:~ toby$ date|sed -e 's/[^0-9]/ /g'
            26 14 48 39     2004
    

    (update) DV:

    Sure I come from an ISP provided address range (free.fr) but the domain is legit, I even have a working reverse DNS.
    That's a common administrative measure. You may also find (as I recently did with sympatico.ca) that the ISP blocks SMTP from subscribers except to its own SMTP server. The solution, in both cases, is to deliver mail via the ISP's SMTP server. It's simple configuring a gateway's sendmail or postfix to do so.
  • 26 Sep 2004 (updated 26 Sep 2004 at 02:07 UTC) »
    raph, thanks for the informative post. I think I'll have to read it twice and follow the links.

    the perturbation of the middle knot damps out very quickly as you traverse (S-shaped) Cornu spiral segments, but doesn't damp out at all with the circular arcs
    IKARUS' circle fitting method is actually well damped, and behaves much more like the second example (as does METAFONT's IIRC).

    I'm willing to tolerate a significantly larger number of primitives in this internal representation than either input through the UI or exported into a font file format after all manipulations have been done
    Doesn't this have some major implications for hinting? I get the impression from FontFocus that stroke-oriented hinting is still a favoured approach.

    I keep mentioning IKARUS because for many years it was the gold standard for these purposes... it's interesting you mention auto-tracing. URW produced a rather nice companion program which traced bitmaps into IK format, called LINUS. I still have a copy of the Mac version.

    24 Sep 2004 (updated 24 Sep 2004 at 05:15 UTC) »
    elanthis: Point taken. The battle for freedom doesn't put me in a joking mood. I'll celebrate when the likes of $CO and M$ are properly buried with a silver stake perforating their carcasses in the necessary manner.

    (Yeah, I know, the stake doesn't have to be silver. But you can't be too careful. And it's 1.17am.)

    23 Sep 2004 (updated 23 Sep 2004 at 17:50 UTC) »
    yeupou: I agree. Could GPL-hate be self-hate?
    22 Sep 2004 (updated 9 Oct 2004 at 16:48 UTC) »
    Cubic Beziers are actually pretty good at making an S curve, in my opinion. Certainly, it's not uncommon in professional fonts to see a single Bezier traverse from a vertical tangent in the upper left corner to another vertical tangent in the lower right.

    I agree Beziers are efficient, but I don't think this should seduce us into concluding they are natural.

    If you draw the main slope of the S with, say, eight arcs, then the centers should describe a fairly continuous path in and of themselves. ... But if you don't do a good job visualizing the radii of curvature, and put the tool in novice hands, my guess is that you'll see bad curves.

    That's a good point. But rather than having the user think about the circles, I wonder if they could use the abstraction of a digital spline and ducks (the non-mathematical meaning of spline). I suspect this is what Dr Karow had in mind when he came up with IKARUS' curve model.

    it's probably not anywhere nearly as painful as you might think to work directly in the space of the Cornu parameters
    That sounds interesting.

    having an uneven baseline or x-height line is a pretty serious artifact
    Surely only subpixel placement can help with non-integral lineskips? But it seems to me that (at a cost of performance, i.e. glyph caching), FontFocus could be adapted to subpixel positioning? (In the vertical, to solve the baseline problem.)

    The CoolType samples in your paper do not work. In my opinion the subpixel rendering on OS X and XP (and Adobe Reader 6) does a fine job; the screen dump can only work in exceedingly rare circumstances (1:1, digital LCD of correct triplet ordering, etc) and should be deleted from the paper as a misleading comparison.

    19 Sep 2004 (updated 19 Sep 2004 at 15:39 UTC) »
    yeupou:
    she said windows, not Microsoft Windows. Well, as I understand it, the company does not own the word windows, but the trade mark Microsoft Windows.
    You missed Michael's point. Regardless of what M$ "owns", in common usage, the word "windows" has gained another (less useful) meaning. In real life this overloading easily results in the kind of ambiguity that his anecdote illustrated.

    109 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!