Satya

Netscaler-based content-switcher with Shibboleth support

Sun, 19 Jul 2009 15:14:36 GMT

Notes on setting up a content-switching Citrix Netscaler with Shibboleth support. I use Netscaler 9, I think.

Assume two servers, S1 and S2, that I want to present as www.example.com. They have Shibboleth. I want SSL and non-SSL both.

Set up Shibboleth Service Providers on both servers. The Host can be S1 and S2, doesn't matter. Set the HandlerURL to /S1Shibboleth.sso or something, rather than /Shibboleth.sso. Some way to distinguish them is needed.

On the Netscaler device, set up S1 and S2 as servers/services with their real internal IP. Call these s1-int and s2-int. Type HTTP, port 80. Set up similar ones called s1-int-ssl and s2-int-ssl with type SSL, port 443.

Set up load balancers, of types and port HTTP/80 and SSL/443, called s1-lb, s2-lb, s1-lb-ssl, s2-lb-ssl. Their IPs can be private 10-subnet IPs. The s1 IP is same for both SSL and non-SSL, and the s2 IP is also the same (but different from s1's). Example: set s1-lb and s1-lb-ssl as 10.0.0.4, and set s2-lb and s2-lb-ssl as 10.0.0.5. Add the services s1-int s1-int-ssl etc. to the appropriate load balancers. You'll need SSL certificates to associate with each one.

Set up content switching policies for your apps on s1 and s2. (See Netscaler manual.) Set up policies for the /S1Shibboleth.sso and /S2Shibboleth.sso as well!

Set up two content switchers: www-ext and www-ext-ssl. Add all the policies to each one, the targets being the appropriate load balancers. That should do it. I don't think you need to set up Apache on the boxes any different. Using S1 as the ServerName should be fine.

You'll need SSL certificates for the SSL vservers. See the Netscaler manual.

Cloned from Satya's website (post comment)

Elaborate security is not better security

Wed, 15 Jul 2009 15:00:44 GMT

I've been trying to set up an online account to access a mutual fund brokerage.

The registration page asks for my social and account number, among a few other unimportant details. Strike zero point five.

Then they deliver the next set of instructions through a web-based "secure" system called zixmail. Third-party, strike one. And haven't they heard of PGP, GPG, and so forth?

Okay. The "email" contains instructions for first login, which do NOT tell me where to go. Anti-phising, or just lame? Anyway, the web site has at least 3 different logins, each slightly different. I'm not even sure I'm using the right one (mutual funds, brokerage, or managed investment? Who knows.)

The login name involves my last name and social. The password involves my mother's name and my date of birth. Strike two. Naturally, I use slightly different names on each site. It doesn't work. I call to have it reset. It doesn't work. They reset again. It works.

At least that was the initial password. I'm asked to change it, and new password must be between 8-12 characters, CANNOT contain special characters (strike 2.5). The instructions include "not contain any special characters. For example, AA-BB-CC". I didn't know what to make of that.

After passing that hurdle, it asked for security questions four. I could pick a question each, out of 4 different lists (one per security question). I happened to pick two questions that had the same answer. Bzzt. The rules say: 1. All four questions must be answered. 2. All four answers must be unique. 3. Answers must contain minimum of 3 alphanumeric characters. 4. Answers must contain only alphanumeric characters and spaces. So I had violated rule 2. Sigh.

After getting through all that, they wanted me to set up a personalized security passphrase and an image from their image library. (The security phrase must be between 3-50 characters.) This is for anti-phishing purposes.

Now is all that stuff actually going to protect my account? Maybe. Better than just username+password, plus a one-time pad that they snail-mail to me? Heck, no. (The US Treasury does that one-time pad thing. It's cool, but then they fail with their random-order on-screen keyboard thing.)

Go go gadget Schneier. Debunk this!

Cloned from Satya's website (post comment)

Rails and Sybase

Mon, 06 Jul 2009 20:33:09 GMT

I just spent 4 hours trying to get Ruby on Rails to talk to Sybase. This is Rails 2.1-ish on Ubuntu 9.04, installed from the debian system (not gems).

It wouldn't work.

No matter what.

I tried installing the sybase adapter:

  gem install activerecord-sybase-adapter -s http://gems.rubyonrails.org

Nope. And that's after freetds was already installed. It wasn't a freetds error -- it kept yelling about activerecord versions, just like odbc below.

I tried the odbc driver:

    gem install activerecord-odbc-adapter

It kept pulling in activerecord 2.3.2 (remember, I have 2.1.0 from Ubuntu's debs). And then borking on mismatched versions.

Note that those two gem commands work. The breakage happens when you try to actually access something, such as User.find(:first) from script/console:

Gem::Exception: can't activate activerecord (>= 2.0.2, runtime), already
activated activerecord-2.1.0

This is because those gems pull in a gem version of activerecord, which is 2.3.2. The apt-get version is 2.1.0. So I ripped out rails and installed it as a gem:

sudo apt-get remove rails
sudo gem install -V rails activerecord-odbc-adapter odbc-rails
# (takes a long time)
sudo apt-get autoremove
sudo apt-get install irb libdbd-odbc-ruby

(You'll need ruby-dev if you want to install a gem like 'mysql', which I do. What, you think Sybase is my database of choice? Hah! Oh wait, the mysql gem won't install. WTF? Oh, ok, I can apt-get install libdbd-mysql-ruby, and that gets me a find() on a mysql database. If that hadn't worked, I'd be yelling.)

I stuck this in ~/odbc and ran "odbcinst -i -s -d ~/odbc" (I have tdsodbc and unixodbc, dunno how much of that is required):

[d]
Description     = d server, name elided
Driver      = /usr/lib/odbc/libtdsodbc.so
Server=fqdn.example.com
Port=4100
TDS Version      = 5.0
[a]
Description     = a server, name elided
Driver      = /usr/lib/odbc/libtdsodbc.so
Server=fqdn2.example.com
Port=4100
TDS Version      = 5.0

Test with this command: isql -v d user pass where d is the 'd' from the odbc file, and user and pass are the actual ones.

Oh, and you have to add /var/lib/gems/*/bin to your path. * is 1.8 in my case. And remove the symlinks from my vendor/ directory, and maybe regenerate a few files with "rails .", as I was doing this to an existing project.

Update: Formatting!

Update: Added the Gem::Esception error as google-bait.

Cloned from Satya's website (post comment)

Ruby on Rails instance variables explained

Sat, 13 Jun 2009 04:25:05 GMT

Instance variables (@foo, for example) are a Ruby concept, not Rails. But it's a little confusing in the Rails' Model-View-Controller paradigm.

Consider that in pure Ruby, an instance variable is available to an object (which is *instantiated* from a class). The instance variable can be used and changed by any methods of the object. This is unlike non-@ variables, which are limited to the scope where they're defined:

def ex1
   foo=1 # defined within the ex1 method
end

def ex2
    if @foo==1
        bar=1
    end
    bar # nil, because it's out of scope
end

def ex3
    @foo=1 # remains defined after ex3 returns
end

Now here's my point: in Rails, instance variables defined in a controller method are available to the view, but not to any models called from that method. Why?

Because a Rails view is NOT a separate class. It's a template and it's part of the current controller object. A model is a separate class.

That's all. It's a simple point, but it can be confusing at first to those not familiar with Ruby or Object Oriented Programming.

Cloned from Satya's website (post comment)

ActiveRecord::find() in Ruby on Rails

Sat, 13 Jun 2009 04:21:29 GMT

When I first started with Ruby on Rails, it took me some time to figure this out. ActiveRecord's find() method will return two different kinds of objects when you call it with :first versus :all.

find(:first) returns a single object. find(:all) returns an array. If no records are found (perhaps because conditions didn't match), find(:first) returns nil, while find(:all) returns the empty array. find(:all, :limit => 1) returns an array, perhaps with a single object in it. find(:first, :limit => 1) or even :limit => 2, returns a single object (or nil).

The result of find(:first) can be checked with the nil? method, and the result of find(:all) can be checked with the nil? as well as empty? method (but nil? should return false if the find() actually was executed).

In perl terms, find(:first) always returns a scalar, find(:all) always returns a list.

Cloned from Satya's website (post comment)

Shibboleth 2 on Ubuntu 9.04

Fri, 05 Jun 2009 18:52:28 GMT

I recently tried to install Shibboleth as a Service Provider on Ubuntu 9.04. Shibboleth 1.3 is End-of-lifed June 2010, so the shibboleth-users mailing list advised me -- strongly -- to use Shibboleth 2. Well, Ubuntu 9.04 doesn't have the packages for it. Debian Lenny does. So here's how you get a Shibboleth 2 SP (Service Provider) on Ubunt 9.04:

Remove libapache2-mod-shib and auto-loaded packages, if you had them installed. Then, install some of the packages required by the shib2 module:

aptitude install libsaml2 unixodbc opensaml2-schemas xmltooling-schemas

and dependencies. The aptitude command will take care of the dependencies.

Download and install the following. These are the shib2 package from Debian lenny (current stable).

http://ftp.us.debian.org/debian/pool/main/s/shibboleth-sp2/shibboleth-sp2-schemas_2.1.dfsg1-2_all.deb
http://ftp.us.debian.org/debian/pool/main/s/shibboleth-sp2/libshibsp1_2.0.dfsg1-4_i386.deb
http://ftp.us.debian.org/debian/pool/main/s/shibboleth-sp2/libapache2-mod-shib2_2.0.dfsg1-4_i386.deb

Or go find them yourself, since the particular versions listed are sure to be obsolete after this article is posted. You want shibboleth-sp2-schemas libshibsp1 libapache2-mod-shib2, in that order. Download the .deb files and run

dpkg -i shibboleth-sp2-schemas*.deb libshibsp1*.deb libapache2-mod-shib2*.deb

Cloned from Satya's website (post comment)

Ruby on Rails and option_groups

Tue, 02 Jun 2009 02:53:14 GMT

Ruby on Rails rocks my socks again. I had a need for a drop-down with options from many "question sets". I did this in the controller:

@disclosure_questions=QuestionSet.find(:all,
        :conditions => ['is_screener=?', false], 
        :include => [:questions])

So QuestionSet was my outer collection, and each one has many questions, hence the include.

Then in my view I did this:

<select name="disclosure_id">
< option_groups_from_collection_for_select(
        @disclosure_questions, 'questions', 'title', 'id', 'number')
%>
<select>

Which says, make option groups from the disclosure_questions, which is a collection of QuestionSet. The sub-groups are composed of questions, which is a method called on each QuestionSet (by virtue of has_many). title is an attribute of QuestionSet, and id and number are attribute of each question.

More help at the Rails API site

Cloned from Satya's website (post comment)

Pidgin logs archiver

Fri, 15 May 2009 23:56:12 GMT

I wrote a little script to archive my pidgin logs in a sepaarte directory. 4 lines, but took 45 minutes to write. Use at own risk. This should be under the Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License. But should this script do anything, or not do something, don't hold me responsible. In fact, don't use this at all.

YEAR=2008
DEST=/home/whatever/chatlogs/$YEAR

# Copy the directory structure.
find . -type d -exec mkdir -p "$DEST/{}" \;

# Copy the files starting with the year.
find . -name $YEAR*.txt -exec mv "{}" $DEST/{} \;

# For the rmdirs, we could use -exec command + instead of -exec command ; but we don't because
# this script simply doesn't run very often.

# Remove any extra directories we made, go depth-first
find $DEST -depth -type d -exec rmdir --ignore-fail-on-non-empty {} \;

# Remove any empty directories in case we took away all the files, go depth-first
find . -depth -type d -exec rmdir --ignore-fail-on-non-empty {} \;

It should not be run from the .purple/logs directory, or wherever. After that's done, you can not go tar cz (create, compress with gzip) the directory.

Cloned from Satya's website (post comment)

Ruby on Rails auto_link bug

Sat, 09 May 2009 15:45:56 GMT

In Ruby on Rails, the auto_link helper takes a string and automatically links URLs and email addresses in the string, producing output suitable for a web page. Today we had a URL that ended in a query string, with a slash in the string, like so: .../end?tag=/mom

auto_link wouldn't recognize the slash as part of the link and ended the link at the "tag=" part. After some googling, I located action_view/helpers/text_helper.rb and added the slash before the dash in the regexp for the query string. This is in a constant called AUTO_LINK_RE.

Update: Swivel blog has a better way to fix this, with an initializer.

Cloned from Satya's website (post comment)

Javascript programming

Thu, 30 Apr 2009 01:40:42 GMT

I spent most of last weekend programming in Javascript (or ECMAscript). I don't ordinarily do that, but I do have a couple of Firefox extensions and a couple of Greasemonkey scripts to my name. I was writing scripts for the web-based Pardus game. It's pretty heavy (for me) Javascript programming and DOM (Document Object Model) manipulation. All the stuff I just mentioned is on my Pardus Auto-mapper page. Usually I just do Ruby on Rails with some light Prototype/Scriptaculous javascripting. But this is about writing apps with just the client (Firefox).

Cloned from Satya's website (post comment)

Custom library path in Ruby

Thu, 09 Apr 2009 00:36:55 GMT

To add a custom or additional library path to a Ruby program: $LOAD_PATH.unshift File.dirname(__FILE__) + '/../lib'

This "unshifts" onto the LOAD_PATH list, the full path to the lib directory, relative to the current file (which is the ruby program). Current file's location is guven by calling dirname on __FILE__.

This was useful for me when I was trying to get railroad to run from my home directory. (Railroad is a utility to graph Ruby on Rails' model/controller relations, very useful, I recommend it.)

Cloned from Satya's website (post comment)

The Internet on South Park

Thu, 09 Apr 2009 00:33:38 GMT

I just watched the South Park episode "Over logging". In it, they showed "The Internet" to be a giant Linksys router (the blue plastic SOHO type, such as WRT54G, with wireless -- I'm a dork, yes). It was a lot like Stargate, underground bunker and all. They should have also had a caption saying "This is what many people actually believe", because this *is* what many people actually believe. "I mean, have you ever *seen* The Internet?" They flash a similar caption when telling the story of the Book of Mormon and of Sci-- I mean, the Super Adventure Club. Oh come on, it would have been funny.

Cloned from Satya's website (post comment)

Pro tip: fatherly advice

Wed, 25 Feb 2009 00:05:29 GMT

My father gave me two pieces of advice, and one skill:

Always be presentable,
Use the right tool for the job, and
Riding a bicycle

Always be presentable

He really meant, "Shave!", but it also applies to general appearance: you never know what kind of opportunities will present themselves when you're wearing ratty shorts and a torn-up shirt. He didn't mean wander around with a suit and tie at all times, of course. This comes in useful, in the literal sense, when you suddenly get a meeting scheduled and... oh, good, you've shaved today!

Use the right tool for the job

Literally, he meant don't use a flat-blade screwdriver for turning a Philips-head screw, but he also meant in general. You can't approach a problem with a mind-set that doesn't fit a problem. I've been struggling with a particular construction problem, that I just fixed (or vastly simplified, we'll find out after the... stuff... dries) by going out and getting the right $5 tool.

Riding a bicycle

He forced me to learn to ride a bicycle, which became my primary transport for years in college and the first few years of work. Everyone should learn to ride a bicycle, and swim. Then you can compete in triathlons (he doesn't say that, I'm saying that). Yeah, no deeper lesson here, except: learn to ride.

The examples I use are literal, because those are the contexts we encontered them in. He's a technical/mechanical person, an electronics guy, so that's what we usually talk about. Soldering, hanging drywall, and stuff. Except we didn't have drywall.

Cloned from Satya's website (post comment)

Making PDFs with JasperReports and Rails 1

Sun, 22 Feb 2009 01:01:01 GMT

Here's part 1 of how to produce a nice PDF in a Ruby on Rails application using JasperReports, a Java code library.

Overview:

The development:

Set up your Rails application to produce XML for the data to be filled into the template.
Make the PDF template using iReport. iReport produces a .jasper file.

The execution:

Your Rails application produces XML.
A small Java program puts the XML and the template together using the JasperReports libraries. It returns the PDF.
Your Rails application accepts the PDF and stores or serves it.

Setting up the template and building the XML should usually be done at the same time. Since I'm not doing a full tutorial on JasprReports, I'll cover things briefly and without screenshots. I'll mostly cover the stuff that's specific to using an XML data source.

First, get your Rails app to somehow return an XML file. Stick a method in the model to build the xml, call the method something like build_xml(). This will take a list of ids of the records that you want in the PDF. I give it a list of ids or an array of grade report data structures because I usually produce a grade report for a given student. By giving this method multiple student ids or data structures, I can produce a whole set of reports in a single PDF file.

By giving it an array of data, each element being one student's grade data, the caller is agreeing to fetch the data from the database. By giving it ids, the builder methods will fetch the data as well as building XML. I do something like this: I put a method in Student class to get the data, and I call that method, and send what it returns to build_xml.

def self.build_xml(grades)
    buffer=""
    xm=Builder::XmlMarkup.new()
    xm.instruct! 
    buffer += xm.grades do
        grades.each do |c|
            build_single_xml(xm, c)
        end
    end
    return buffer
end

build_xml declares a buffer where the XML data will accumulate. xm is the builder object. instruct! puts in the declaration that this is an XML document. xm.grades produces a "grades" XML element, and the block inside it produces the contents of that element. We loop over the array of "grades", calling build_single_xml for each. build_single_xml builds each student's grade report data. So, our XML is going to look like:

<grades>
    <student id="">
        ...data...
    </student>
    ...repeat for each student...
</grades>

Notice that each student's id is included as an attribute.

I won't cover how to generate all the XML at this point. Let's turn to the JasperReports template... in part2.

Cloned from Satya's website (post comment)

Craigslist scams with a laptop

Thu, 19 Feb 2009 02:51:35 GMT

I recently posted a laptop for sale on craigslist. I got several responses right away, but none of the ones I contacted got back to me. Here are some of the responses. I most cases I answered the questions asked or asked them when they'd like to come look at it. The more scammy-sounding ones:

Do you still have these item for sale please let me know? thanks

Hello Seller is this item available if so i will like to know the condition and i am willing to pay u by Pay Pal i will be waiting to hear from u as soon as possible

I also got one offer for half the price I listed, and a couple of emails asking me to call a (local) phone number. These were legitimate, I think. Lots of them asked if it was still available (and nothing else). A couple wanted to know about battery life and the OS. All of them got answers.

Another asked for a couple of details and did reply to me when I responded, though they declined to buy. I thanked them for at least having the courtesy to communicate properly. Oh, here's a beauty:

Thanks for your reply and am glad to know that it is still available for sale.However i have some few questions on the products your selling.
1,I would like to know the reason you are selling it..?
2,What is the current condition of the item..?
3,How much would you take last for it, though am okay with the listed price..?
4,Do you have the original box for it and the receipt..?
5,Would you please allow me to pay through paypal even though i would cover the shipping because am currently not local at the moment,Am participating in a research project and i would like to purchase this item from you on behalf of my colleague whose computer crashed.

If you do not have a PayPal account, you can do so by going to www.PayPal.com and set up your account it is free,easy and guarantees transaction safety.Once you send me your paypal email account, I will send you the payment for your item as well as $100 USD to cover shipping vai USPS EXPRESS MAIL INTL.
I would be glad if my request is favorably consider.
Thanks

Shipping? *International*? No thanks. Here's another one that at first just asked if I have it, and after I said yes, came back with this:

Nice hearing back from you, I would be happy to purchase it from you. I'm shipping this international to my friend, and I think it would be less costly if we could ship it straight from you to him. I'm ready to pay you the total amount of $250,for the item and $100 the shipping fees ,via USPS EXPRESS MAIL INTERNATIONAL (EMS) and i will pay you via paypal if that would be acceptable. Please reply and include your paypal email address if you are interested. Thank you and God Bless you.....

The "total amount" of $250 is wrong, it's more than what I had posted. That's in addition to the other things wrong with this message.

A few days later I re-posted the laptop. This time I said explicitly that I want cash, and local pickup, and no exceptions. Almost immediately I got a couple of emails asking if I still had it. Then I got this gem:

I am very glad to hear back from you. I am a University Senior lecturer residing in [New York]. I came across this ad on Craigslist and thinking of my Son's Birthday coming up soon, I would love to get an awesome present for him, which he really wants, he was currently transferred from USA to West Africa with his team on a research on Human development under world Health Organization.I'll be paying you through PayPal,it's secure and protects two parties in a transaction. I will forward my son's residential address to you for shipping as soon as the payment reaches you. Please kindly get back to me ASAP,so that i can make the payment there.
NB: I will be paying you $200 for item and i will include $100 to cover up the cost price for the shipping fee. Get back to me with your PayPal email asap.

Regard,

Sent from my iPhone

And then this one. I never mentioned anything about being "loaded":

INTERESTED IN YOUR DELL LAPTOP. WHAT DO YOU MEAN EXACTLY LOADED?? REASON FOR SELLING?? ANYTHING WRONG?? SLOW??

And this one:

I am [name deleted] by name and will like to buy your item as a birthday gift which will be commencing very soon and will like to buy it which i will highly elated want you to help me to ship it to her directly over there which i will be responsible for the shipping cost via Usps Expres mail service,so i will be glad to pay you securely via Paypal or via Cash money order which will be deliver to your at your doorstep in cash,so i will be offering you $350 for it including the shipping cost via Usps express mail as mentioned above and if this is okay by you,you can send me your paypal email address with your full name as appear on your account and if you feel you prefer cash money order for immediate delivery...you can mail me the details below...

Your full name:
Your home address as you will like to be the receiver...
Hope to hear back from you soon.
God Bless.....

Okay, dude. I am not going to ship this laptop anywhere, especially not when you're offering me such juicy bait: nearly double the price I quoted.

I eventually got a couple of responses that were local, and sold it to the first one I got.

Cloned from Satya's website (post comment)

To-do lists and food plans

Sat, 14 Feb 2009 15:24:55 GMT

Based on some stuff I posted on identi.ca, and that JaredW. Smith also posted, here are a couple of lifehacks.

(Lifehack: an adjustment to your habit that makes life easier.)

I maintain a spreadsheet with a daily plan of what I'm going to have for lunch and dinner. It's not in any great detail, just two entries every day, for about a week in advance, each entry being something like "pasta" or "KCJI" or "leftover chinese". That helps me plan what to buy, keeps me from too quickly repeating, lets me track where I ate and when, and helps prevent situations where various different leftovers are piling up in the fridge. This plan should be flexible, though, so you can switch stuff around based on immediate needs and wants.

I have a TODO list on my desktop. A program called tomboy, which is available in at least Debian and Ubuntu, lets me put short sticky-notes style windows on my desktop. Tomboy is actually a personal desktop wiki, and there are other sticky note programs out there. A simple text file may work better for some.

The key is, to-do list should contain items to do *today*, and other immediate stuff. Otheriwse stuff just stays on it forever. "Exercise" isn't a TODO list item. "laundry", "treadmill", "clean oven" are TODO list items. Those are the "things I need to do today". Sure, you can also have things like "fix that bug in that program", but if you don't get to it soon, remove it from this list. And yes, you can use a personal bug-tracking system for this, as well.

Cloned from Satya's website (post comment)

Did I go to Wando?

Mon, 09 Feb 2009 22:15:36 GMT

A 10-second-long transaction at the bank, and the girl at the counter asks me, did I go to Wando (high school)? No? A brother? Any family member? Which is weird, because I'm not from around here! (And it's fairly obvious.)

I've had a lot of strange questions, but that's a first.

Cloned from Satya's website (post comment)

Charleston County Recycling news

Mon, 02 Feb 2009 23:28:04 GMT

I received the Charleston County recycling newsletter last week, and it brought up some questions. I emailed them and they answered -- every single question was answered, in full, in a reasonable manner. Now that's service.

First, plastic window envelopes. You can rip out the plastic and recycle the paper parts. Yay!

Then, paper with confidential information: You can shred and drop off the shredded paper at a drop-site.

So yeah, as long as the contamination is kept to a minimum, they'll accept paper in any form.

#1 and #2 plastics are only taken in bottle form because that's what the vendors want. Oh, and they don't take recyclables #3-#7 because there isn't a market for it. Shipping it elsewhere would just raise our taxes, and we won't stand for that, right?

Cloned from Satya's website (post comment)

Yahoo domain names fiasco

Sat, 31 Jan 2009 17:26:52 GMT

I have a domain at Yahoo Small Business that expires next month. Since Yahoo now charges about $30 or more per year, I decided to transfer the domain to Dotster.

I always set up whois privacy for my domains, so that a Whois database lookup (this is public information) returns a placeholder postal address, email address, and phone and fax numbers instead of my own. Yahoo and Dotster both claim that they will forward anything received at these points of contact to me, as long as it doesn't look spammy. Yahoo contracts with MelbourneIT for this, by the way.

Unfortunately, the contact information that Yahoo puts into the Whois system isn't tagged. Everyone using the Whois privacy "feature" gets the same details. How does Yahoo distinguish between them? It doesn't. When I requested a domain transfer via Dotster, Dotster sent an email to the administrative contact, which was Yahoo's placeholder address. I never saw that email. Only by turning the Whois privacy off, and asking Dotster to re-send the request, did I get Dotster's email. In the meantime my private information is exposed to the world of domain criminals.

Dotster's private registration *does* use tagging, the email address listed is something random. So they can forward real requests to me. I have tested this successfully.

In the meantime, I did a Whois lookup for a domain I had cancelled several months ago. To my surprise and dismay, Whois showed my personal information for this cancelled domain! The technical contact was listed as MelbourneIT, so I contacted them and asked to have my information removed immediately. They asked me to contact Yahoo. Yahoo asked me to call them. I did, and had the information removed. Apparently, at Yahoo Small Business, "cancelling" your domain does NOT mean "de-registering"! Horrible. I'll never be using Yahoo Small Business again. I'll be avoiding anything by MelbourneIT, too.

I'm not the only one with this problem.

Cloned from Satya's website (post comment)

BoingBoing article on pneumatic tubes

Thu, 22 Jan 2009 17:20:05 GMT

I saw this title on Boing Boing "Pneumatic tube-based postal systems of the late 19th century", but first I saw the word "portal" instead of "postal" and thought they were making some Portal(the game)-related joke. Too bad they weren't.

Insert your own intertubes joke here.

Cloned from Satya's website (post comment)

Sansa E200P playlist

Wed, 07 Jan 2009 04:29:59 GMT

I have a Sansa Sandisk E200P music device that shows up as a USB mass storage device under Linux. After much digging around, I found a couple of web sites and eventually realised that the playlist format is "extended M3U" (and these are DOS ASCII files, acording to a forum post). So I wrote a script that will build playlists. You will need perl and the ability to edit simple perl code. Sansa playlist maker

Cloned from Satya's website (post comment)

Kino downgrade

Mon, 05 Jan 2009 05:49:34 GMT

I use kino for video editing. The newest version which comes with Ubuntu Intrepid Ibex (8.10), version 1.3.1, has some bugs related to scene splitting. After trying (not very hard) and failing to get the kino devs to acknowledge the bug, I tried to compile kino on my box. Mistake. First I had to apt-get install about 100 packages (really, I counted and there were definitely over 100). Then it complained that frame.h was including libavcodec/avcodec.h and a couple of others, which couldn't be found. Some googling turned up a laugh at Redhat

"Why is it that every decent multimedia production tool always requires stuff to build its tarball that is not part of the fedora repositories?"

and then this, which said to reference avcodec.h and so on, i.e. remove the libavcodec/ directory qualifier from the #include lines in frame.h However, the compiled kino segfaulted when I tried to run it. Then it occurred to me to go grab an *older* debian package, one which I know works -- the one from Ubuntu Hardy (8.04)! See, so far I'd been trying to find *newer* packages. I added Hardy main to the apt sources.list file, and got synaptic to "Force version" for kino to Hardy, then locked kino's version. Too bad it still screws up the scene splits, though. After pinning ("force version" is also known as "pinning") various packages, I could not fix the problem. Eventually I tried editing a recently-captured video file. It kept the scene splits. So, apparently, my old files have some issue that makes them lose the scene split data after a kino FX operation. Weird. This confusion arose because I've only edited old files since the Ubuntu upgrade. Sigh. What a waste of time.

Cloned from Satya's website (post comment)

Recycling packaging material

Thu, 01 Jan 2009 16:25:53 GMT

Based on a post at Go Green Charleston blog, I had some thoughts:

I got a large number of packages this year. I freecycled all the boxes and packing material (including peanuts and brown paper) to a family that's moving to New Jersey. There were 2 small air bubble bags that went with it. The best part: I bought two new computer cases which came with those custom tall, thin, boxes. Well, this family has two computers they're moving, so those boxes and their foam padding came in handy!

Now the treadmill I bought has a ton of waste, where can I recycle it?

several plastic bags with no recycling number (the biggest problem)
bits of foam packing material (trash, a few pieces saved for later re-use)
plastic ties (trash)
metal bits (can these be recycled somewhere in Charleston county?)
cardboard (recyclable)
wood! (where can I recycle this, short of leaving it outside to decompose?)

Update: Plastic bags went to the grocery store. I think they know how to sort them out. The metal bits will be taken to a recycling drop-site, and are in my car now. The cardboard was recycled, and I still have the wood. I *could* build a shelf, but why?

Cloned from Satya's website (post comment)

Build notes for the Windows/Game system

Wed, 24 Dec 2008 15:51:13 GMT

Objective: a sub-$200 (in new parts) gaming computer capable of playing a not-too-demanding modern game like Battlefield 2 without a lot of lag.

Pictures are here

AMD Sempron 64 3800+ Manila 2.2GHz Socket AM2 Single-Core Processor Model SDA3800CNBOX - Retail $29.99
Kingston 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Dual Channel Kit Desktop Memory Model KVR800D2K2/2GR - Retail $22.99
Antec NSK4480B Black 0.8mm cold-rolled steel construction ATX Mid Tower Computer Case 380W Power Supply - Retail $69.99
JetWay JM26GT4 AM2 NVIDIA GeForce 6100 Micro ATX AMD Motherboard - Retail $39.99

Subtotal $162.96
Tax $0.00
UPS 3 DAYS $35.56
Order Total $198.52

I'll be using the old 80GB IDE hard drive in this one, with a 20GB Windows partition and the rest for data. I'll also be using an nVidia GeForce 6600LE that I bought for the other computer and didn't use (since the other one had an HDMI-to-DVI adapter). The 6600 is supposed to be better than the 6100, and anyway the on-board one supposedly uses system RAM instead of dedicated VRAM.

The AMD CPU is about $40-$50 cheaper than a comparable Intel CPU. My alternative was the Intel Wolfdale E5200 at $82, but it is dual-core unlike this AMD Sempron. That dictated the choice of motherboard, which is simply the cheapest one with AMD support and a PATA connection. I want analog video out (VGA) anyway, because of the way I have the shared monitor hooked up.

2GB for the RAM, because 1 isn't enough and Windows XP is supposed to choke on anything more than 3. Since the board and RAM are dual-channel, 2GB it is.

With the IDE hard drive and IDE DVD drive, I made sure to get a case where all the drives mount in the same direction. But I still had to use the long cable I bought for the other build. Apparently it works with everything but the DVD burner I have, and I need DVD burning in my primary Linux box, and I don't need it in my game box. So the game box has a DVD-ROM drive in it and seems to work with the long PATA cable. Antec supposedly makes good cases, so there we go, it's the cheapest Antec case that fulfills my needs.

I will install the CPU and RAM on the motherboard first, before opening the case. But I will need to open the case to ground myself.

The case opens at the top with 2 thumbscrews. Sharp-ish edges as noted in the reviews, but come on, it's a steel case. Be careful. The side panels lift up and out, and even the front panel is fairly easy to remove as long as you can pull the panel while releasing each of the 6 tabs, tightest one first in lockpicker-style.

The case contains the usual spare hardware, screws and so on, and a power cable. It has a rear 120mm fan with speed control switch, a CPU duct, and no other fans besides the PSU (power supply). There is a bracket for a front 80mm fan, and the ability to mount one under the side duct.

Cute. There's a loose piece of 8-inch cable tie in there. There's a plastic wrap-tie in the screws pouch, along with a case badge and an unexplained metal bracket.

With this 380W PSU, I have to watch my power consumption. I'm not sticking anything complex in there. Besides the inventory above, it's going to be a hard drive, an optical drive, and an nVidia GeForce 6600.

The hard drive cage is detachable, which is good because there are no drive trays. It's hard to screw in a hard drive while supporting it with one hand inside a cramped box. Eww, the 5.25" optical drive bays have pop-off metal covers. At least the gap left by them will be covered by the front panels more-modern re-attachable plastic covers.

Isn't this nice: the front plate bezel pops off completely. I can stick an air filter in there, but who's going to clean it, and when? The case has several air holes. I'd like to cover those with filters, but again who's going to clean/change them? Hmm, front panel comes apart into two pieces. Handy.

Motherboard box comes with the usual manual, driver disk, back plate, 2 SATA and 1 PATA cables, and the board. This board takes 24pin ATX power (with the last 4 pins covered by a removeable tab, what's up with that?) plus 4-pin power connection, plus CPU power connection.

I have broken the seal on the CPU box. As usual with current CPU retail boxes, it's filled mostly with heatsink and manual. I hope it comes pre-greased, as I don't have any thermal grease. It does come pre-greased. Adding the HSF (heatsink-fan combo) is a little easier on this than on the Intel board, maybe because I'm doing it before inserting into the case.

The RAM was a little hard to get out of it's packaging, but it's in there now. It's uncomfortably close to the CPU, and I'm worried about heat. AMD CPUs have historically run pretty hot. The Northbridge heatsink is also pretty close to the CPU.

Case feels pretty light with all the panels off. The stock back panel (over the rear mobo connectors) is wedged in real good, had to hammer on it with the handle of a large pair of scissors. Unlike the other Zotac board, this mobo one has just 6 screw holes for attaching to the case.

Okay, the metal grilles over the 5.25" bays are practically impossible to break off by hand. Only the top bay is open and given that this is an all-PATA system with a board that has only 1 PATA connector, I need to make sure the IDE cables will fit with the optical drive in the top bay. I don't want to use the extra-long cable I have because I suspect it has issues.

It's not easy to detach the hard drive cage as there are two screws under it, in an awkward position even with the case laying on its side and not full of dust. The cage does have metal tabs, so you don't have to use one hand to awkwardly hold up the drive while using a second to hold the screw and a third to turn the screwdriver. There are holes in the right-side of the case for hard drive screw access. The hard drive bays have silicone grommets, and the long screws necessary are included with the case. A long magnetic screwdriver is also needed to get in from the right.

When I am ruler of the universe all PCI and similar cards will have handles to pull them out.

Oh, neat, the rear slot covers pull out. I thought they would have to be broken off, since they lack screws. But they're just held in place by friction. With the video card installed, I notice that the large heatsink on the GPU (Graphics Processing Unit) and VRAM (Video RAM) is below the card when the case is in its operating position. Anyone else notice that heat rises? I guess it's supposed to be more of a radiator and spreader than a convective cooler, but still.

I knew I did something stupid: badge on front of case has been glued upside down.

Due to the CPU duct and the way the side panel closes (it slides down from above by about half an inch) it's a little hard to close, especially with my big IDE cable in there. The duct helps move the cables away from the CPU and the video card, though, but I should route them better. Which I have now done. The side panel closes a little easier now. Sure, the fan wires are taped to the bottom of the PSU, but who cares.

I've added a few air filter swatches inside the side panel grille and in the CPU duct.

This power supply, an Antec Earthwatts 380W, also has a hard switch on the back. Powering up....

Well, that's interesting. My ancient Windows XP boot CD does not recognize the on-board network adapter. Oh, well. Either it'll be on the motherboard's disc or I can find a generic one online and transfer visa USB. Here's a hint, though, my XP disc may not work on new computers much longer. Or are modern CPUs still capable of emulating 8086 and Pentium?

Interesting, XP detected the network right after installing the drivers from the mobo disc. No reboot needed. But I don't think it's doing very well with the video -- dragging a window is slow and jerky. Bet it doesn't have the drivers. (Yes, installing the drivers made it much better.)

How bad is it to reverse the polarity of the electron flow across a photon emitter? Translation: I plugged in the power LED backwards

Downloaded the nVidia drivers. 80 megabytes, sigh. Installing now. My standard loadout for a Windows system includes Firefox, AVG Free, and Zonealarm free, and I'm adding WinSCP to this system. I also installed a motherboard monitoring app from the mobo disc, as well as sound, network, and System Management Bus (SM Bus) drivers. Eventually the Windows Device Manager stopped showing yellow exclamations or red whatevers.

Also downloaded the latest Quicktime, because Septerra Core keeps crashing on startup. I wonder what else is missing or unsupported.

New system doesn't feel very fast, despite turning off all the eyecandy stuff that Windows brings in (the equivalent of turning off Compiz, I guess). Battlefield 2 seems to run better.

According to the mobo monitor, system temperature is about 85 degrees Fahrenheit and CPU varies about 90-96 Fahrenheit. According to my wattmeter it takes 60 to 90 watts and about 750mA.

Cloned from Satya's website (post comment)

CD-ROM on ubuntu Ibex

Sat, 20 Dec 2008 04:16:35 GMT

Google-bait: On Ubuntu Ibex 8.10, with an IDE (PATA) CDROM/CD-RW drive, the drive is detected just fine, dvd+rw-mediainfo sees it, but on trying to mount it with Gnome I get messages like "Unable to mount media" and other things implying that there is no disc in the drive, as if the disc is not sensed. Adding this to /etc/fstab solves the problem:

/dev/scd0 /media/cdrom0 udf,iso9660 user,noauto,exec,utf8 0 0

Cloned from Satya's website (post comment)