73 lines
2.8 KiB
HTML
73 lines
2.8 KiB
HTML
|
{% extends "_layout.html" %}
|
||
|
|
{% block title %}I2P Status Notes for 2006-05-09{% endblock %}
|
||
|
|
{% block content %}<h3>I2P Status Notes for 2006-05-09</h3>
|
||
|
|
<pre>-----BEGIN PGP SIGNED MESSAGE-----
|
||
|
|
Hash: SHA1
|
||
|
|
|
||
|
|
Hi y'all, tuesday rolls 'round once again
|
||
|
|
|
||
|
|
* Index
|
||
|
|
1) Net status and 0.6.1.18
|
||
|
|
2) baz
|
||
|
|
3) ???
|
||
|
|
|
||
|
|
* 1) Net status and 0.6.1.18
|
||
|
|
|
||
|
|
After another week of testing and tweaking, we pushed out a new
|
||
|
|
release earlier this afternoon which should get us into a more
|
||
|
|
stable environment to make improvements from. We probably won't
|
||
|
|
see much effect until it gets widely deployed though, so we may have
|
||
|
|
to wait a few days to see how it goes, but measurements will of
|
||
|
|
course continue.
|
||
|
|
|
||
|
|
One aspect of the latest builds and releases that zzz brought up
|
||
|
|
the other day was that increasing the number of backup tunnels now
|
||
|
|
can have a substantial impact when done while reducing the number of
|
||
|
|
parallel tunnels. We don't build new leases until we have a
|
||
|
|
sufficient number of live tunnels, so the backup tunnels can be
|
||
|
|
quickly deployed in case of a live tunnel failure, reducing the
|
||
|
|
frequency of a client being without an active lease. This is just a
|
||
|
|
tweak to a symptom though, and the latest release should help
|
||
|
|
address the root cause.
|
||
|
|
|
||
|
|
* 2) baz
|
||
|
|
|
||
|
|
"baz", the new machine bar donated has finally arrived, an amd64
|
||
|
|
turion laptop (with winxp on the boot disk, and a few other OSes in
|
||
|
|
the offing through the external drives). I've been working it over
|
||
|
|
for the last few days too, trying to test a few deployment ideas on
|
||
|
|
it. One problem I'm running into though is getting gcj working on
|
||
|
|
windows. More specifically, a gcj with a modern gnu/classpath. The
|
||
|
|
word on the street is pretty negative though - it can be built
|
||
|
|
either natively in mingw or cross compiled from linux, but it has
|
||
|
|
problems like segfaulting whenever an exception crosses a dll
|
||
|
|
boundary. So, for instance, if java.io.File (located in libgcj.dll)
|
||
|
|
throws an exception, if its caught by something in net.i2p.*
|
||
|
|
(located in libi2p.dll or i2p.exe), *poof*, there goes the app.
|
||
|
|
|
||
|
|
Yeah, so its not looking too hot. The gcj folks would be very
|
||
|
|
interested if someone could jump in and help out with the win32
|
||
|
|
development, but viable support doesn't seem to be anything
|
||
|
|
imminent. So, looks like we'll have to plan on continuing to use a
|
||
|
|
sun jvm on windows, while supporting gcj/kaffe/sun/ibm/etc on *nix.
|
||
|
|
I suppose thats not really that bad though, since its the *nix users
|
||
|
|
who have problems packaging and distributing jvms.
|
||
|
|
|
||
|
|
* 3) ???
|
||
|
|
|
||
|
|
Ok, I'm already late for the meeting, so I should wrap this up and
|
||
|
|
tab over to the irc window, I suppose... see you in a few ;)
|
||
|
|
|
||
|
|
=jr
|
||
|
|
-----BEGIN PGP SIGNATURE-----
|
||
|
|
Version: GnuPG v1.4.2.2 (GNU/Linux)
|
||
|
|
|
||
|
|
iD8DBQFEYPdizgi8JTPcjUkRAnL/AJ9CwQ0EsmYxfOY2uFznQYKm4xMS3wCfXkan
|
||
|
|
VDUPoPbUGi8bVf/h7Ao6jtY=
|
||
|
|
=qDIQ
|
||
|
|
-----END PGP SIGNATURE-----
|
||
|
|
|
||
|
|
|
||
|
|
</pre>
|
||
|
|
{% endblock %}
|