Pete On Software

RSS Feed

C# 6 Features – Expression-bodied Members

I have finally gotten around to looking into C# 6. It was released in July 2015 and available starting in .Net Framework 4.6, but I just hadn’t had any occasion to use it. In order to “force myself” to dig into it a little more, I decided to write a series of blog posts about the new features. The first one out of the gate is Expression-bodied Members.

Expression-bodied Members are just a syntactic sugar on the language. I’m not angry about things that are syntactic sugar. In fact, one of my favorite things about C# 3 was automatic properties, so I’m all in. In reality, it is just about expressing intent more clearly and more succinctly. Here is a very simple example of a method that uses this new feature.

In this case, I’ve defined a public method that takes an integer and returns an integer. As we are reading left-to-right, there is nothing weird yet. However, after the method name declaration, we change it up and throw the “Anders Operator” aka “Hash Rocket” aka “LINQ Symbol” aka “Goes To” aka “=>” into the mess. Just like LINQ, you can put an expression right afterwards and that is what is evaluated. We are missing the curly braces, sure, but we are also missing a return statement. You don’t need it.

You can also use this simple syntax for properties. Take a look at the following class:

The FullName property is now just a read-only expression. The alternative would have looked like this:

Even if you one-lined the FullName property, it would not be as clean as the Expression-bodied version

Fantastic. We’ve seen a way to clean up our code a bit, no down side, right? Well, of course there is a down side. This is only really useful and clean-looking if you have a very simple one-liner to express. If you ramp this guy up a bit, it starts to break down.

This doesn’t even compile, for probably very obvious reasons. You can’t do a bunch of work on separate lines and then just hope the last evaluation is what is returned. The error you get is “Invalid token ‘=’ in class, struct, or interface member declaration”. The reason being that C# thinks that the method is just public int DoLotsOfStuff(int input) => input += 2; After that, it is trying to figure out how input = input * 2; makes sense as a free standing line within the class. That isn’t how you declare a method, a property, or a field, so C# freaks out. Of course you could jam all of that on one line, but it may start to get confusing the more you try to do.

So, Expression-bodied Members can be a nice clean way to express your code simply. However, the danger here is that developers could try to play “code golf” to get their code to fit on one line. That would really decrease readability, which defeats the entire purpose. So, I feel like this is a feature to use carefully and sparingly.

Podcast Episode 43 – Older Programmers

Old ManWhy is the average age of a developer so much less than the average age of a professional worker in the United States? Is it because the industry as a whole is striving to keep people out? Are companies replacing people as soon as their salary starts to get higher with people who make half as much and will work twice as many hours for the same salary? Check out what I think about it in this week’s episode.

Links Mentioned in this Show:
Developers Fear Age 30
Older Workers are More Knowledgeable, but Harder to Find
Where do all the Old Programmers Go?
Silicon Valley’s Dark Secret
Neural Network in the Browser

You can also subscribe to the podcast at any of these places:
iTunes Link RSS Feed Listen on Stitcher
In addition, my podcast is available on DoubleTwist and the Windows Marketplace.
Thanks to all the people who listen, and a special thanks to those who have rated me. I really appreciate it.
Listen here:

NullIf

NullIfThis post is about a handy little function that works across many database systems, including Sql Server, Oracle, and MySql. I often forget about it and at times even end up coding some workaround that would have been a lot easier if I had just used NullIf(). NullIf takes two parameters. The first parameter is the value to check and the second parameter is the value that should equate to NULL. Let’s take a look at an example to hopefully make this clearer.

First, we will create a table to work from, populate it with some data, and select the results out so that we can see what we have visually.

Our table's contents

In this case, the Street Address column is nullable, but for Keith’s record the developer chose to insert a blank space instead of a NULL for a missing address. If we try to just write a query to get rid of the nulls, we still have the blank space issue. Here is a query that uses COALESCE to get rid of the NULLS and its results.

Our results with COALESCE Only

You see that we still have the blank address to deal with. One work around is to use a case statement. Something like “CASE WHEN COALESCE(StreetAddress, ‘Not Provided’) = ” THEN ‘Not Provided’ ELSE COALESCE(StreetAddress, ‘Not Provided’) END AS StreetAddress”. But, that is quite a mouthful and we repeat ourselves several times. However, if we could get COALESCE to treat a blank address like a NULL, we’d have been in business in the first place. That is what NULLIF() does. It evaluates the value we specify as a NULL (in this case a blank space) and then the rest of the query can treat it like a NULL. Here is that example

Our results with COALESCE and NULLIF

That’s much more succinct and I feel like it conveys our intent much more easily. Here’s another practical use case for NULLIF(). Given the following table and values, we want to find the average sale on each given day.

We might write something like the following

However, when you do, you get a divide-by-zero error. Sometimes in more complicated situations, that can be a bear to track down exactly what is evaluating to zero in an equation. However, if you ask the denominator to evaluate as NULL whenever it is 0, SQL is much happier. Here’s our query now:

Now, we easily get our results.

Results of Dividing by NULL

This helps identify which records are causing our issue and can now easily be dealt with. If we don’t like NULL in our results, we can use ISNULL() or COALESCE() to put a sensible default answer. Done and done.

Podcast Episode 42 – The Answer to Life, the Universe, and Everything, just kidding – MS Build 2016

MS Build 2016In episode 42, I don’t have the answer to Life, the Universe, and Everything, but I do have thoughts and reactions to Microsoft’s 2016 Build conference. I cover Bash on Windows, Ink, Bots, and Free Xamarin for everyone!


Links Mentioned in this Show:
XKCD Thing Explainer
Cleartext

You can also subscribe to the podcast at any of these places:
iTunes Link RSS Feed Listen on Stitcher
In addition, my podcast is available on DoubleTwist and the Windows Marketplace.
Thanks to all the people who listen, and a special thanks to those who have rated me. I really appreciate it.
Listen here:

SQL Pivot – Static or Dynamic Columns

Pivot ArrowThis is going to be another one of those posts that will surely help some people, but it is far more likely to help me later 😉 I have culled these techniques from a variety of sources across the Internet, but I’m not entirely sure who taught me what in order to give proper credit.

If you want to follow along, I went to the Microsoft Sql Product Sample Site and downloaded “Adventure Works 2014 Full Database Backup.zip” and unzipped and restored it onto my database server. That gives me the OLTP version (rather than the OLAP) of the 2014 Adventure Works database. To start off, let’s take a look at the following query:

This returns results like these:
Initial Month-based Results

However, what if you wanted to easily see how each of those customers’ sales looked month over month in comparison to other customers? Typically, a data analyst might take a result set like that and use a tool like Microsoft Excel to create a pivot table. That would let your rows be per customer and your columns would be per month. Reporting wonks could also use a tool like SSRS to pivot the data for presentation to users. However, if you are using Sql Server 2005 or later, you could also use the PIVOT command.

If your data that you’d like to pivot on has a finite and known quantity, this is very easy. I’m going to change the query above to show how to do that:

A few peculiar things to notice. First, the values that become the new columns are not quoted. They are treated like literals. I didn’t have to say ‘January’, ‘February’, etc. Secondly, notice that I had to name the pivot table (as pvt here). Even though I don’t reference it, if I omit it, the Sql statement won’t parse and execute. Look at the results below. Notice a third strange thing. Even though I selected out all of the columns (using SELECT *), it did not actually return all columns like you would normally see in a join situation. It showed the AccountNumber column from the SalesData CTE, but then it ignored the OrderMonthNames and TotalSales columns from SalesData. The PIVOT was smart enough to know that those values were being represented in other ways in our new dataset.

Pivoted Month Results

If you want/need to get rid of the NULLS in the results, you can do it the same way that you would do it in a regular query. I could change the “SELECT *” to be “SELECT AccountNumber, ISNULL(January, 0.00) AS January, ISNULL(February, 0.00) AS February” and just continue that pattern for each month.

Simple enough and that gets us pretty far along the way. But, what if I wanted to instead see customer sales by product category? I could see every product category that exists in the system right now and do a static pivot. However, everytime someone updated the system with a new product category, we would have to modify our query (potentially many queries across the enterprise). Is there a way to do that?

No, not directly.

However, we can use Dynamic Sql to dynamically determine our static columns, if that makes sense 😉 Let’s take a look at the query needed to just return the standard results, grouped by customer and product category.

This returns the following results:
Initial Dynamic Pivot Results

There are 8 unique category – subcategory combinations in these results, but it doesn’t take much imagination to see how new products can be added all the time. Probably several new subcategories a year. How can we handle that? The key is those columns listed out in the PIVOT operator definition. Let’s see how we can build that dynamically using a temp table to make working with the data easier.

The bulk of the work here is just being done in this line

That just select a distinct list of categories and builds a comma delimited string out of them. The value in that variable looks like this:
[Bikes-Road Bikes],[Clothing-Socks],[Clothing-Jerseys],[Clothing-Caps],[Components-Mountain Frames],[Accessories-Helmets],[Bikes-Mountain Bikes],[Components-Road Frames]

Because of the spacing in the categories, that’s why we used QUOTENAME. That makes sure the name will work in SQL statements. Ultimately, the executed query looks like this:

You can see how that is now very much like our initial static example with the months. The query still has the NULL problem, though, so to fix that you can create another variable so the SELECTed columns can contain ISNULL but the PIVOTed columns will not. That looks like this:

Basically, I just made the distinct list of category names to be a distinct list of category names wrapped in ISNULL. Then I used the @SelectColumns variable in the select statement and the @PivotColumns variable in the pivot list. The resulting Sql looks like this:

And the resulting data looks like this:
Dynamic Pivot Final Results

Hopefully you find this helpful (I’m sure future me will be back here). If you have any questions, let me know in the comments.

« Older Entries