<?xml version="1.0"?>
<rss version="2.0.">
  <channel>
    <title>Advogato blog for ncm</title>
    <link>http://www.advogato.org/person/ncm/</link>
    <description>Advogato blog for ncm</description>
    <language>en-us</language>
    <generator>mod_virgule</generator>
    <pubDate>Thu, 24 Jul 2008 20:26:40 GMT</pubDate>
    <item>
      <pubDate>Wed, 16 Jul 2008 22:09:49 GMT</pubDate>
      <title>16 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=278</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=278</guid>
      <description>&lt;a href="http://advogato.org/person/chalst/diary/193.html" &gt;chalst&lt;/a&gt;: My experience is that the account deletion&#xD;
threshold is set so (artificially) high that the only way to&#xD;
get a spammer account deleted is to mark them down as&#xD;
spammers unless it's dead clear they're not.  Even then lots&#xD;
of fake accounts get through.  Mostly the real spammers&#xD;
don't start spamming immediately; they appear to be&#xD;
accumulating fake accounts and holding them in reserve.&#xD;
&#xD;
&lt;p&gt; Anybody who gets their account deleted can just sign in&#xD;
again and behave less spammerishly next time, so there's no&#xD;
reason to use kid gloves.</description>
    </item>
    <item>
      <pubDate>Mon, 14 Jul 2008 21:12:31 GMT</pubDate>
      <title>14 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=277</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=277</guid>
      <description>I have altered the "s_t" and "c_t" ligatures in my copy of&#xD;
the Linux Libertine font so the loops are higher.  They look&#xD;
really good in Firefox, when it uses them.  (I haven't&#xD;
discovered yet how Firefox -- really, libpango -- decides&#xD;
whether to substitute ligatures; sometimes it does,&#xD;
sometimes not, sometimes both in one paragraph, e.g. this&#xD;
one.) Now I need&#xD;
to fix the italic, bold, and bold-italic versions, and get&#xD;
them all accepted upstream.  I'd still like to find a better&#xD;
way to get them used by default.</description>
    </item>
    <item>
      <pubDate>Thu, 10 Jul 2008 22:34:33 GMT</pubDate>
      <title>10 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=276</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=276</guid>
      <description>&lt;a href="http://advogato.org/person/laburu/diary/6.html" &gt;labaru&lt;/a&gt;: What?  I'm one of &lt;a href="http://advogato.org/person/fzort/" &gt;Mauro&lt;/a&gt;'s&#xD;
biggest admirers.  Sincerely: Go, Mauro!</description>
    </item>
    <item>
      <pubDate>Thu, 10 Jul 2008 21:42:20 GMT</pubDate>
      <title>10 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=275</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=275</guid>
      <description>&lt;a href="http://advogato.org/person/redi/diary/146.html" &gt;redi&lt;/a&gt;:&#xD;
No kidding! "We do not use C++ exceptions."  What&#xD;
a way to dispel the myth of a Google populated by geniuses.&#xD;
 Oh, and StudlyCapsTypeNames, too.  If it's true that "taste&#xD;
is the feminine of genius", they've shot both.  Oh, look,&#xD;
enumerator names that get clobbered by common vendor header&#xD;
macros?  Fail.  No rule against public virtual functions?&#xD;
Fail.  No requirement to qualify member accesses with&#xD;
"this-&amp;gt;"?  Fail.  "Don't use an unsigned type", without a&#xD;
peep about how signed overflow is undefined behavior?  Fail. </description>
    </item>
    <item>
      <pubDate>Thu, 3 Jul 2008 03:18:36 GMT</pubDate>
      <title>3 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=274</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=274</guid>
      <description>I just changed my linux-libertine typeface to make Firefox&#xD;
use the looped ligatures for "st" and "ct".  I don't know&#xD;
what would be the "right" way to do it; I just added those&#xD;
ligatures to the default set.  Unfortunately, they're kind&#xD;
of ugly (not loopy enough), and there's no ligature for&#xD;
"sc".  To fix.</description>
    </item>
    <item>
      <pubDate>Thu, 3 Jul 2008 02:28:42 GMT</pubDate>
      <title>3 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=273</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=273</guid>
      <description>&lt;b&gt;Solar&lt;/b&gt;&#xD;
&lt;p&gt;Whenever I read a serious discussion of the prospects of&#xD;
solar power in the U.S., I read that huge expanses of&#xD;
land would be needed, presumed to be in faraway deserts, to&#xD;
cover with solar panels or mirrors.  I look out the&#xD;
window, though, and see huge parking lots all around.  These&#xD;
lots are mostly uncovered, but not because people don't want&#xD;
to park in covered spaces.  Rather, it would cost money to&#xD;
roof them over.  Roof them with solar power collectors,&#xD;
everybody benefits: the lot owner gets another income&#xD;
stream, the car owners get covered parking, and the power&#xD;
company avoids transmission costs and losses.&#xD;
&lt;p&gt;The same argument applies to sports stadiums.  Warehouse&#xD;
and mall roofs would last much longer if shaded, and their&#xD;
owners' cooling bills would be lower.  Many bridges would&#xD;
look better, and be safer in bad weather, covered; likewise&#xD;
freeways.  It seems to me there is plenty of public and&#xD;
semi-public land that would benefit from another use.</description>
    </item>
    <item>
      <pubDate>Tue, 1 Jul 2008 23:15:03 GMT</pubDate>
      <title>1 Jul 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=272</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=272</guid>
      <description>The Debian Iceweasel/Firefox 3 crash (tickled by the&#xD;
linux-libertine 2.8.14 release) is fixed today with&#xD;
libpango1.0-0 1.20.4-1.  (Yay Sebastian Dr&amp;ouml;ge!)  But why is&#xD;
it "urgency=low"?</description>
    </item>
    <item>
      <pubDate>Thu, 26 Jun 2008 21:00:49 GMT</pubDate>
      <title>26 Jun 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=271</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=271</guid>
      <description>&lt;a href="http://www.advogato.org/person/fzort/diary/117.html" &gt;fzort&lt;/a&gt;: I would hate for you to think me your a.o&#xD;
nemesis.  I enjoy, and learn from, your postings.  That one&#xD;
just tickled four peeves at once.  Emulating a class&#xD;
hierarchy in C, though, looks like make-work.&#xD;
</description>
    </item>
    <item>
      <pubDate>Thu, 26 Jun 2008 00:39:07 GMT</pubDate>
      <title>26 Jun 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=270</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=270</guid>
      <description>&lt;a href="http://www.advogato.org/person/fzort/diary/116.html" &gt;fzort&lt;/a&gt;:&#xD;
Your mistake was fourfold.  First, using&#xD;
StudlyCaps names rots the brain.  Second, even thinking&#xD;
about Singleton rots the brain.  Third, the visible&#xD;
presence of keyword "delete" usually indicates bad design.&#xD;
Finally, class hierarchies are an optimization step; first&#xD;
code without.  The biggest difference between writing a&#xD;
program well in C++ and in "straight C" is in how&#xD;
expressively powerful are the libraries you have access to.&#xD;
&lt;p&gt; The place where you code very differently in C++ vs. C&#xD;
is in libraries.  Of course any big enough program will have&#xD;
parts that increasingly resemble libraries, but&#xD;
programs don't start out big.&#xD;
</description>
    </item>
    <item>
      <pubDate>Wed, 25 Jun 2008 22:02:55 GMT</pubDate>
      <title>25 Jun 2008</title>
      <link>http://www.advogato.org/person/ncm/diary.html?start=269</link>
      <guid>http://www.advogato.org/person/ncm/diary.html?start=269</guid>
      <description>Sure enough, downgrading linux-libertine from 2.8.14 back to&#xD;
2.7 appears to have fixed the Debian Iceweasel/Firefox&#xD;
crashiness.  UPDATE: Turns out this is known, Debian&#xD;
#488015. UPDATE 2: ...and #487979, both today, but from&#xD;
Germany, where they have a nine-hour jump on me.  (Yay Sven&#xD;
Joachim, excellent bug reporting.)  It's a good thing there&#xD;
are other browsers.</description>
    </item>
  </channel>
</rss>
