dz>0

I lost a Z key on a 12″ iBook G4 and need a replacement.

Tonight was the first class, and I very much enjoyed the lecture. Dr. Leonid Perlovsky is quite engaging, clearly enjoying and thinking about his topic while taking the measure of the audience and responding as his algorithms would - with emotional intelligence and dynamic modification.

A long description of the course is here.

The text is Neural Networks and Intellect: Using Model-Based Concepts

I’m trying to think up some project to play with the ideas presented.
A long loop back cognitive tape machine playing Pink Floyd comes to mind.

Some of my notes from the class-
Perlovsky considers culture to be strongly influenced by language. He specifically calls out the successes and problems of the US as being largely due to the English language.
At some point he wrote a paper on how the brain could use Quantum Mechanics in its operation, but it was not clear that he still considers this to be likely.
He considers evolvable ontologies to be a worthwhile current challenge.
Recommends Rhetoric for Alexander as good reading.

I’ve seen this claim before, that Eclipse is slow, but I had not experienced the issue quite so personally before. Most of my Eclipse development to date had been small components to other projects, or in one case a project that might have been about one thousand files. Now I am trying to turn a project that comprises some 40,000 files into an Eclipse project(s) and it is not going as I hoped. Subclipse takes tens of minutes in some operations, the Binary Search (CDT?) that gets started on Eclipse launch takes about 5 minutes. And for other reasons, the integrated gdb view does not seem to be working with the executable because of a time out during launch; though this could be a simple configuration problem. In any case, it looks like I’ll have to spend much more time breaking the project into small bits, which should be done anyway, I had just hoped to start building and debugging before I had to do house keeping.

I know that Subversion can handle the project size, and I know that make is suitable for the level of complexity involved in building everything, I just want to put it all in a nice integrated environment for those of us that would prefer not to use gdb from the command line, or who would like to use the refactoring tools in Eclipse rather than grep and sed!

I signed up for an IEEE short course because it sounded interesting, Bio-inspired and Cognitive Algorithms for Recognition, Data Mining, Tracking, Fusion, Prediction and Language Understanding.

Well see how much fun this turns out to be.

I had some trouble getting VNC to connect to my CentOS box. The firewall prevented the connection.

The fix is simple enough:
[root]#/sbin/iptables -I RH-Firewall-1-INPUT -m state --state NEW -p tcp --destination-port 5901 -j ACCEPT

Saving the fix so it remains on boot seems a little more challenging; naturally I am mixed between the command line and graphical. I would like to know why my VNC connection seems to hang sometimes.

[root@sv ~]# apropos vnc
Xvnc (1) - the X VNC server
vncconfig (1) - configure and control a VNC server
vncpasswd (1) - change a VNC password
vncserver (1) - start or stop a VNC server
x0vncserver (1) - VNC server which continuously polls an X display

Prefer vncserver to Xvnc

Seems like the desktop from my vnc server gets hosed after a few months. I assume it is gnome, but I would think I get a new gnome session with each login. vncserver -kill :1 did the trick.

vncserver :1 -name G5 -depth 16 -geometry 1400x1000

At work we have a large project. Possibly the most complex I’ve been involved in with respect to organizational issues. A number of groups are involved, all trying to build the main software product, as well as integrate their own software. Also, the product in development has a number of dependencies that really need to be managed in an automated way because we can not depend on each vendor to download the correct version of all the tools in the tool chain. In fact I really do not trust them to download, configure, build, and install gcc correctly.

So the question is, can I use Eclipse in a platform independent way to manage dependencies that include the gcc compiler, gdb, xerces, libitt(?), and the project code itself?

Of course it needs to work with svn and not actually depend on each vendor being willing to use Eclipse and CDT.

What I really want for myself is to be able to use Eclipse for development and debugging, but to support the standard:
./configure
make
make install

I may do some experiments at home to see if I can get Eclipse to work as I desire.
It needs to work fast as well - I notice at work that Eclipse and Subeclipse plug-in seem to be really slow right now, and I do not know if this is because of the project organization, or if it can be fixed at all.

I need to get mod_jk working so I can connect jsp to my apache server. The foolishness I’ve used involving Proxypass is just not going to work any longer.

I’ve been using the apache VirtualHost feature to allow for some experimentation without having to worry about breaking parts of my website that actually work. The key here is that Joker.com now allows the use of wildcard * in the DNS setting, so any *.drewk.net gets sent to my webhost. It used to be that I had to add each blog.drewk, test.drewk, dx.drewk to everything I did, and the updates took as long as 24 hours. Now, I just have the wildcard, so I can add any virtual host name I like and go.

< VirtualHost *:80 >
DocumentRoot /xxx/yyy/zzz/nnn/drewk
ServerName 192.168.8.123
ServerSignature off
DirectoryIndex index.html index.htm index.shtml
< Location >
Order deny,allow
deny from all
allow from 192.168.8.210
DAV On  DavDepthInfinity on
< /Location >
< /VirtualHost *>

The above little bit finally allows me to use WebDAV with Dreamweaver and connect to my server without any worry about exposing WebDAV to the unwashed public.

There is an open meet in Woonsocket, RI on the 19th, so I’m keeping my weight below 94Kg and trying to keep a good program between meets.

I expect to use Retrospect Client to backup Linux - it will be interesting to see if SELinux causes me any particular difficulties. Nope, it didn’t.
Installed the 7.5 client, but I can not see it from my Mac. Probably an iptables issue. Nice to see that man retroclient provides some info.

Retrospect tries to connect to the retrospect client TCP port 497
/sbin/iptables -I RH-Firewall-1-INPUT -m state --state NEW -p tcp --destination-port 497 -j ACCEPT

/sbin/iptables -I RH-Firewall-1-INPUT -m state --state NEW -p udp --destination-port 497 -j ACCEPT

Not sure which one did the trick, but Retrospect can now see the client.

Looks like I’m able to back up my Linux box now. Of course it is only really useful if I prove to myself that I can recover necessary data.

(more…)