The Daily Parker

Politics, Weather, Photography, and the Dog

Lotus 1-2-3 epoch day weirdness

The ancient spreadsheet package Lotus 1-2-3 set "0 January 1900" as its day zero. Whenever you entered a date into a Lotus spreadsheet, the program actually stored the number of days before or since that mythical date. Microsoft Excel needed to maintain compatability with Lotus early on, so it set 30 December 1899 as its day zero, which worked very well except for dates between 30 December 1899 and 1 January 1900, and it added the other mythical date 29 February 1900 because Lotus had that bug as well.

Flash forward a few decades, and today brings us a little bit of weirdness that only shows up in the American Month-Day-Year format.

Today is 4/21/15 in the U.S., and also day 42115 in Excel.

This bit of nerdiness brought to you by the Daily Water Cooler and one of The Daily Parker's readers.

Things I didn't read while pulling apart an Include block

...and also preparing for a fundraiser at which I'm performing tomorrow:

And did I mention Apollo After Hours?

Still debugging

I was here until 7:30 last night and would probably stay that late tonight if I didn't have a prior commitment. At least last night I got to see this:

At least I've isolated the code causing the problem. Unfortunately it's one of the most-called methods in the application. Sigh.

Debug mode

I was going to write a couple of entries today on various topics, but then this happened in production:

That's a production database getting overloaded. I am now going to continue debugging it.

To-do list lengthening, day remaining same size

In the reading queue:

Did I mention that DUKE WON?!

More stuff to read on the plane

With a little more than five days until my next international flight, I'm stocking up my Kindle:

UAT release this afternoon. Back to the galley.

Logic says this is highly improbable

If you have a set of unit tests for a large application, and not every one of them passes, then you don't know whether the application works correctly. In fact, failing unit tests are evidence that it isn't working.

Also, marking them with the [Ignore] attribute doesn't count as fixing them.

Also, deleting the part of the unit test that fails doesn't count either. An empty test method is not a passing test unless the method being tested is also empty. And an empty test method should not ever be committed to the source repository.

</rant>

Oddly-sized photos

You may have noticed that the photos I've posted lately have unusual aspect ratios and sizes. This is because I have been lazy.

Usually, I upload photos from my camera or phone to my laptop, process them with Adobe Lightroom, and crop them to a pleasing 2:3 aspect ratio. Lately, though, I've just shot them through Google Hangouts from my phone. I have little understanding of Google's choices but they seem to be around the byte count and not around the dimensions.

Here, for example, is a shot of O'Hare Terminal 3 from this morning:

Google did that one at 1236 x 695, at 249 kB. The original is much larger (4 MB), but the same aspect ratio.

I may re-edit the photos later, after landing.

Update: here's the same photo after going through Lightroom:

Sad day in software development

After 38 years of publication, Dr. Dobbs is shutting down:

Why would a well-known site, dearly loved by its readers and coming off a year of record page views, be sunset by its owner?

In one word, revenue. Four years ago, when I came to Dr. Dobb's, we had healthy profits and revenue, almost all of it from advertising. Despite our excellent growth on the editorial side, our revenue declined such that today it's barely 30% of what it was when I started. While some of this drop is undoubtedly due to turnover in our sales staff, even if the staff had been stable and executed perfectly, revenue would be much the same and future prospects would surely point to upcoming losses. This is because in the last 18 months, there has been a marked shift in how vendors value website advertising. They've come to realize that website ads tend to be less effective than they once were.

So rather than continue with Dr. Dobb's until it actually loses money, [our owners] decided to sunset the site — a sudden end to remarkably robust and wondrous journey that began 38 years ago.

In the 1980s and 1990s, Dr. Dobb's was the best source of information about C/C++ programming, bar none. I've read the magazine and the website off and on for about 30 years. I'm sad to see it go.