Running out of entropy

Some of our services make a million UUIDs per day. I have recently been worried about how random they are, and especially with the question are we creating duplicates? I use Java's java.util.UUID.randomUUID() to create the UUIDs. These are created using random data from /dev/urandom. I have been told that /dev/urandom never blocks and never runs out of randomness. I find this hard to believe.

You check on how many bytes of entropy are available using

    cat /proc/sys/kernel/random/entropy_avail

Our counts are always less than 200. That is not enough. And so my concern is that I have a system that depends on random UUIDs that does not have enough entropy and so, so as not to block, it is using math to get randomness. I don't trust math randomness.

I don't yet have the full picture yet. Hopefully I will soon.

See Check available entropy in Linux.

Slides for your generic talks

Recently we had the generic TED talk. I think I am going to start collecting the images that seem to be used by all talks. The inaugural image is, without a doubt the most used in more slides than any other, the iceberg.

Searching browser history using Google Chrome search engine

Hum. If you add to Google Chrome the search engine chrome://history/?q=%s you can quickly search your browser URL history by typing Cmd-l h search-text.

Grumble about messaging conversation indistinguishability

Why, why! oh mighty God, do Slack, Skype, Messenger, iMessage, and every other messaging application not make any visual distinction between two or more conversations? Squint at the screen and you will see that every conversation looks identical -- even across messaging applications! How many times have you and I and, most likely, the developers of these applications sent a message to the wrong conversation? Lots. Please, oh pretty please, allow me to at least change the background color of conversations.

$0.43 for Psychotherapist Barbie services

All the wonderful applications and toys that requires an internet service for operation have a very short life. Most application startups don't last two years. That great calendar extension, or small office workflow coordinator, or automatic uploaded video curation you use every day quickly becomes inert. You anxiously wait for their death. The situation for toys is even worse. Almost no toy manufacturer has had to provide service support and maintenance. Once the toy is sold, the only worry a manufacturer had was that of product liability suites. So what are we to expect as we move ever forward into the internet-of-things?

Suppose that Hasbro sells "Psychotherapist Barbie" and she needs to store years of voice recordings and to run complex voice, tone, and sentiment analysis algorithms so as to synthetically voice the appropriate response. How long is Hasbro going to operate this as the number of active customers and maintenance income dwindle?

Hasbro should not run the services. Instead, Hasbro should use the Barbie owner's own Amazon Web Services account to run them on. That is, on behalf of the owner Hasbro will provision the AWS services needed to use Psychotherapist Barbie. Once running, Barbie's owner pays for the services until he or she no longer wants to use the toy. If Barbie's owner prefers Google Cloud Services or Microsoft Azure then Hasbro would provision there instead.

We need manufacturers that sell applications and toys requiring connectivity, storage, and computation for their operation to use the owner's preferred cloud services provider. This is quite easy to do with applications or toys that are used in isolation. For applications or toys that require a network of users then the solution is a little harder as the manufacture would need to implement services that share without centralization and to scale from two users to many, or perhaps millions, of users.

I look forward to the day when I receive my monthly AWS services bill and see listed the applications and toys I use that make my life a little easier and more enjoyable.

PDFs, comments, and Dropbox

I love commenting on long-form documents. If the author put the necessary work into creating a cogent examination of a topic then I am going to put in the effort to understand it. And I do that by making comments. Some are questions. Some are observations. Some are petty, especially the typographic ones.

When you like commenting you want to have the space needed to comment. I seem to recall that I once forced an author to use double line spacing and an extra wide right margin so that I had room to write. That was great. When Microsoft Word added the comment bubble user-interface I did not think my commenting life could be better. I was wrong. Online collaborative commenting is even better.

Most of my documents are written using Google Docs. Docs has weak formatting and content structuring tools, but it does have great commenting (and version control). I am willing to lose a little structure and style for that. Not all documents come my way via Google Docs, unfortunately. Most of them use Adobe's PDF as authors want structure and style over commenting. Especially, it seems, those working in elementary and secondary education. Sigh.

Today I wanted to read and share my comments on RIDE's "Creating a Shared Understanding of Personalized Learning for Rhode Island," a PDF document. So I looked around the web for tools. There are a fair number, but I was happy to find that Dropbox allows for online collaborative commenting of PDFs. Just add the document to Dropbox and share it. Anyone viewing the document sees the comments and anyone wanting to add comments need only login with Dropbox or Google credentials.

Another online problem solved.

3 times a loser

Last night the Town Council for the second time since the November election had to appoint a replacement School Committee member. I was not appointed. I am disappointed; more so than I thought I would be. Losing 3 times in so many months drains the spirit.

Liz Gledhill nominated me and Joe Viele seconded the nomination. Since Scott received the necessary 3 winning votes my nomination was never voted on. Meg Healy, Abel Collins, and Bryant Da Cruz appointed Scott Mueller to the vacated seat.

I want to thank Liz and Joe for having the courage to support my positions and this governance rookie. I also want to thank all those people across this country that stood up for what is right over these last weeks as it was their example that inspired me to try again.

Scott was not a good choice. Healy, Collins, and Da Cruz made a politically safe choice and, moreover, a safe Democratic party choice. The ramification of their choice is that the SC is effectively the same SC as we had before the November election. An election that clearly called for change. Roland Benjamin and Alycia Collins will do their best to shift the SC's focus towards, I am guessing, finer budget control and reducing the barriers that obstruct educating all of our children. The other 5 members and the Superintendent will continue as they always have. The brazen submittal to the TC of a school budget that increases spending by double the State mandated limit clearly shows that they are not adequate to the task of their responsibility. Shame on them. And shame on the TC for not listening to SK's voters.

Universal health-care is great for business

I can tell you what would be great for business in the United States -- universal health-care. Those with great ideas for new businesses can take the risk of losing their savings, but not their lives or those their children. Businesses can compete globally unburdened from the financial and moral sole provision of the health of their employees and their families. Businesses can hire the best from across the country because they are not encumbered by the 50 different state health insurance regulations and limitations. Lastly, and most importantly for those with little hands, little minds, and little care for others, billions of dollars of increased shareholder value.

Let a 1000 flowers bloom and 300 million people breathe easier each and every day.

Too much kit

I have been reading and thinking about Sam Mustafa's new game rules Freejumper, spaceship combat, and Aurelian, the crisis in 3rd century Rome. They are both very interesting and likely make for enjoyable play. They both use game-specific cards for activation and (almost) no dice. In my mind they are both boardgames without a board and with figures. I have nothing against games that use card activation, resource management, or other game mechanics. I really enjoy playing SAGA, for instance, and it is also a boardgame without a board and with figures.

I think what bothers me about these games is that there are too many parts that must be had. To play a dark ages games using the SAGA rules each player must have the faction's "battle board" and their customized dice. To play an ancients game using the Aurelian rules each player needs to have its deck of cards and the unique tokens. These kit are not cheap. For SAGA the rules are $30 and 2 sets of dice are $36 for a total of $66 before playing the first game. For Aurelian the rules are $29 (PDF) and 2 sets of cards are $40 (or, if you print them yourself, $20 in inkjet ink and $10 in sleeves) for a total of $69 (or $59) before playing the first game.

I don't think my aversion is to the cost, our hobby is expensive, but there is ample difference between a high entry cost vs a moderate ongoing cost. Clearly that is where my mind is. I believe that to play a wargame it must be enough to simply have a copy of the rules, figures for both sides, a few rulers, several generic dice, and some generic tokens.

South Kingstown School Committee

I ran for South Kingstown School Committee this past November and so this blog has been quiet as I worked on my campaign. My online campaigning was on Facebook as it gave me the broadest reach into the community and the community an easy way to find me and to keep current. Here is a complete capture, I think/hope, of my postings and comments.

Paint Shaker

To have the itch to paint miniatures again. My paints have been sitting unused for much of a year and many have the pigment separated from the medium. It would take me only a few minutes to shake them all, but this needs a tool! So I built this paint shaker from parts I had on-hand.

It really does not work as well as I had hoped, but it does help. A ball-bearing in the paint container would, I suspect, greatly help stir up the pigment. The kids absconded with all that I had and so I will need to get more.

Bash is useful up to a point

As far as I can tell is there no way in Bash to pass a concatenated array to an internal function and not have the resulting string split apart. Eg

function f() {
    for e in "$@"
        echo $e

A=( x y z )

f a b "${A[@]}" 'c d' "e f"

will output

c d
e f

when I expected it to output

x y z
c d
e f

Google and both failed me, as did trying lots of variations.

Observations on the School Committee's retreat

If you are not following my South Kingstown School Committee candidate Facebook page, but are interested in South Kingstown School Committee's happenings then you might be interested in my observations on the School Committee's retreat this past Friday, July 15: the good, the bad, and the ugly.

“What’s in a name? That which we call a rose by any other name would smell as sweet.” Not in software!

Names have great signification in code. A good name can make the object's value comprehensible and usage clear. A variable called milk_jug is likely to have all the aspects of a jug, but specialized to only hold milk. I had a problem with picking a name.

I had a variable that did not itself hold the value, but acted as though it did. Is was not a proxy. Nor was it a reference. The variable held only a portion of the value but could itself get the remaining portion when needed. This kind of delayed acquisition behavior is common in programming. It is used when the remaining portion of the value is costly to acquire, eg takes a long time to compute or retrieve, or holding it puts a burden on the system, eg using a significant amount of memory.

I ended up calling the variable mik_jug_winding. Winding as in "The river winds through the valley." I like it, but given that I felt the need to write a posting about it perhaps it lacks the needed signification. 

Which Cement Mixer Are You Based On Your Zodiac Sign?

This kind of astrology is important to me. I'm a

Bartell Gas Powered Concrete Mixer
You are a dependable individual, willing to help whether your friend needs a shoulder to cry on, a secret to keep, or cement to be mixed.

Primary and Secondary Tools

Earlier today I was in a UX design meeting and I used the terms "Secondary Tool" and "Primary Tool." (Application and tool are interchangeable terms.) Understanding the difference between these two is very important as their UX needs to be very different. Most application teams think that they are building a primary tool when, I am sorry to inform them, they are not. Very few teams work on primary tools. So what is the difference? I hope the following helps explain this.

Primary Application/Tool

A primary tool is one that the user uses daily. It is one where the user will have many sessions within a short period of time -- a few weeks. This rapid experience of having successes and recovering from failures builds the user's confidence in the his or her ability to use the tool. Moreover, this confidence encourages the user to explore the tool for better means to perform tasks and opportunities to use the tool for unforeseen outcomes. This confidence brings with it the need for far less feedback that an operation has been completed successfully and instead feedback can be limited to erroneous results.

Secondary Application/Tool

A secondary tool is one that is not a primary tool. It is used infrequently -- less than once per week. This infrequency disposes the user to having to relearn the tool's operation at the start of a session. Only the most rudimentary operational knowledge is retained between sessions. The user is most successful when he or she is guided in performing a task from its beginning to its ending. Redundant orientation information such as task milestone displays, activity histories, and possible future work, etc are displayed beside the specific task step at hand. Feedback for success and for failure is always provided.

I am running for School Committee in South Kingstown

I am running for School Committee in South Kingstown, RI.
Let the grumbling begin.

BoardDocs and removing forced page breaks

This posting is mostly to remind myself how to eliminate page breaks from Emerald Data Solutions's BoardDocs. BoardDocs is used by lots of organizations that want control over their governance documents. The South Kingstown School Committee just started using it. Its configuration is such that when the detailed agenda is printed there will be a page break after each agenda item. I don't want this -- either on paper or PDF. To fix this I installed Stylebot and added the following style for BoardDoc URLs to turn off forced page breaks

* {
    break-after: auto;

This seems to work. My preference would be for a less heavy handed solution, however.

SQL and the (newly discovered) OVER operator

I have a tool that takes a range of ids and processes each document with an id within that range. I wanted to run this tool in parallel with each instance processing 100,000 documents. The problem is that document ids are not contiguous. There are gaps. So a simple criteria of document id + offset is not sufficient. I needed an SQL statement that detailed contiguous range of 100,000 document ids.

As with most ostensibly complicated SQL the solution is in creating a virtual table (ie inner select) from an existing table. My virtual table needed to be sorted on document id and a sequence number given for each row. To do that I needed to use this in the column selection of the virtual table.

ROW_NUMBER() OVER( ORDER BY document_id ) - 1 row_number

I had never heard of the OVER operator. It applies a function to a row of an ordered relation. So the ROW_NUMBER() function gives the first row a value 1, the second 2, etc. Since I would be grouping using division I need the row numbers to start from 0 and so the subtraction of 1.

Now that I have row numbers and rows ordered by document id I needed to find within groups of 100,000 rows the minimum and maximum document ids. I used a simple group by row_number / 100,000 to get the row groupings. Then used MIN(document_id) to get the group's minimum document id. Same for maximum document id. The final SQL is as follows and it runs fast enough for the 50M records I needed to use it with

  FLOOR( y.row_number / 100000 ),
  MIN( y.document_id ),
  MAX( y.document_id ),
  COUNT( * )
      ROW_NUMBER() OVER( ORDER BY document_id ) - 1 row_number
      documents x
  ) y
  FLOOR( y.row_number / 100000 )
  MIN( y.document_id )

Now that I have the data I can use a bash script to read the data and produce a series of commands that are feed to xargs for parallel execution.

superglue + baby powder = gap filling putty

I bought some of Games Workshop's liquid green stuff and have been waiting to use it to clean up some plastic kits I am currently working on (well, next in line). Today, however, I found out about the superglue + baby powder putty. Dries rock hard and can be filed to shape.