gsfake hang problem is solved.

We still have some timing-dependent glitches in the regression tests,
though.

1 files changed, 0 insertions, 188 deletions

diff --git a/www/upstream-bugs.html b/www/upstream-bugs.html
index 98b4e188..55f2c9d6 100644
--- a/www/upstream-bugs.html
+++ b/www/upstream-bugs.html
@@ -56,7 +56,6 @@ maintainer, we've tried to kick these upstream.</p>
 <h1>Links to Open User-Visible Problems</h1>
 
 <ul>
-<li><a href="#gpsfake">gpsfake hangs under mysterious circumstances</a></li>
 <li><a href="#bluetooth">Firmware problems in some Bluetooth devices can hang them</a></li>
 <li><a href="#pl2303">Linux pl2303 driver on openwrt 2.4 kernel can hang when device is read at unexpected speed</a></li>
 <li><a href="#pthread_create">pthread_create() fails to return when called in background</a></li>
@@ -87,193 +86,6 @@ with Fedora Core 5 was buggy.</a></li>
 
 <h1>Open User-Visible Problems</h1>
 
-<h2 id="gpsfake">gpsfake hangs under mysterious circumstances</h2>
-
-<p>gpsfake hangs when run with some versions of Python, but the bug is
-not solely Python's; either the underlying kernel or the GCC version
-used to build Python are implicated in some way we don't yet
-understand.</p>
-
-<p>The symptoms are very obvious, the twirling-baton progress
-indicator freezes after the first read-write interaction with the
-slave daemon. We're trying to localize this bug.  Here's what we
-know:</p>
-
-<p>&nbsp;</p>
-
-<table border=1>
-<tr>
-<th>Label</th>
-<th>Distribution</th>
-<th>Kernel</th>
-<th>Python</th>
-<th>Compiler</th>
-<th>Libraries</th>
-<th>Reported-by</th>
-<th>Status</th>
-</tr>
-<tr>
-<td>A1</td>
-<td>Ubuntu 8.04</td>
-<td>2.6.24-21</td>
-<td>2.5.2</td>
-<td>gcc 4.2.3</td>
-<td>glibc 2.7</td>
-<td>esr</td>
-<td>SUCCEEDS</td>
-</tr>
-<tr>
-<td>A2</td>
-<td>Ubuntu 8.10</td>
-<td>2.6.27-7</td>
-<td>2.5.2</td>
-<td>gcc 4.3.2</td>
-<td>glibc 2.8.90</td>
-<td>esr</td>
-<td>FAILS</td>
-</tr>
-<tr>
-<td>A3</td>
-<td>Ubuntu 8.10</td>
-<td>2.6.27-7</td>
-<td>2.4.5</td>
-<td>gcc 4.3.1</td>
-<td>glibc 2.8.90</td>
-<td>esr</td>
-<td>SUCCEEDS</td>
-</tr>
-<tr>
-<td>A4</td>
-<td>Ubuntu 8.10</td>
-<td>2.6.27-7</td>
-<td>2.4.6</td>
-<td>gcc 4.3.2</td>
-<td>glibc 2.8.90</td>
-<td>klein.eric&#x40gmail.com</td>
-<td>SUCCEEDS</td>
-</tr>
-<tr>
-<td>A5</td>
-<td>Ubuntu 8.10</td>
-<td>2.6.27-7</td>
-<td>2.5.2-dbg</td>
-<td>gcc 4.3.2</td>
-<td>glibc 2.8.90</td>
-<td>esr</td>
-<td>FAILS</td>
-</tr>
-<tr>
-<td>B1</td>
-<td>Debian unstable</td>
-<td>2.6.26, 2.6.27</td>
-<td>2.5.2</td>
-<td>gcc 4.3.2 </td>
-<td>glibc 2.7</td>
-<td>bzed</td>
-<td>FAILS</td>
-</tr>
-<tr>
-<td>B2</td>
-<td>Debian unstable</td>
-<td>2.6.27</td>
-<td>2.4.6</td>
-<td>gcc 4.3.2 </td>
-<td>glibc 2.7</td>
-<td>bzed</td>
-<td>SUCCEEDS</td>
-</tr>
-<td>B3</td>
-<td>Debian unstable</td>
-<td>2.6.27</td>
-<td>2.5.3</td>
-<td>gcc 4.2.4 </td>
-<td>glibc 2.7</td>
-<td>bzed</td>
-<td>FAILS</td>
-</tr>
-<tr>
-<td>C1</td>
-<td>Debian etch (chroot)</td>
-<td>2.6.27</td>
-<td>2.5 (release25-maint) </td>
-<td>gcc 4.1.2 (prerelease)</td>
-<td>glibc 2.3.6</td>
-<td>bzed</td>
-<td>SUCCEEDS</td>
-</tr>
-<tr>
-<td>C2</td>
-<td>Debian etch (chroot)</td>
-<td>2.6.27</td>
-<td>2.4.4</td>
-<td>gcc 4.1.2 (prerelease)</td>
-<td>glibc 2.3.6</td>
-<td>bzed</td>
-<td>SUCCEEDS</td>
-</tr>
-<tr>
-<td>C3</td>
-<td>Debian etch (chroot)</td>
-<td>2.6.27</td>
-<td>2.5 (release25-maint)</td>
-<td>gcc 4.1.2 (prerelease)</td>
-<td>glibc 2.7, but compiled against glibc 2.3.6</td>
-<td>bzed</td>
-<td>SUCCEEDS</td>
-</tr>
-<tr>
-<td>D</td>
-<td>Debian experimental</td>
-<td>2.6.27</td>
-<td>2.5.3</td>
-<td>gcc 4.3.2</td>
-<td>glibc 2.7</td>
-<td>bzed</td>
-<td>FAILS</td>
-</tr>
-<tr>
-<tr>
-<td>E</td>
-<td>OpenBSD</td>
-<td>current for 2008-12-30</td>
-<td>2.5.2</td>
-<td>gcc 3.3.5</td>
-<td>libc.so.50.0</td>
-<td>ckuethe</td>
-<td>SUCCEEDS</td>
-</tr>
-</table>
-
-<p>&nbsp;</p>
-
-<p>The compiler is the one reported by Python in its initialization
-banner as its build compiler, not the stock compiler of the
-distribution.</p>
-
-<p>Analysis:</p>
-
-<ul>
-<li><p>The bug appears to require Python 2.5.  Backing down to 2.4
-avoids it in every environment we have tested.</p></li>
-
-<li><p>The bug appears not to be a gpsd problem. The A2 vs. A3 results
-were run on the same machine with the same gpsd binary and instances
-of gsfake/gps.py, a few minutes apart.</p></li>
-
-<li><p>The bug appears not to be a compiler optimization problem. The
-python-2.5-dbg interpreter is built with -O0.</p></li>
-
-<li><p>The bug appears not to be a problem with glibc. The Ubuntu 8.10
-tests all use glibc 2.8.90 but get different results.</p></li>
-
-<li><p>B1 vs B2 means GCC version can't be the key difference
-either. Furthermore, bzed has examined the Python source code and says
-the pty/tty code has not changed between 3.4.6 and 2.5.2 either.</p>
-</ul>
-
-<p>The finger seems to point at something in the core Python
-implementation, possibly having to do with the Global Interpreter Lock.</p>
-
 <h2 id="bluetooth">Firmware problems in some Bluetooth devices can hang them</h2>
 
 <p>The baudrate-hunting code in <code>gpsd</code> tickles a serious