The Bit Bucket

We all use IN when writing queries:

SELECT ProductID, ProductName, Color, Size 
FROM dbo.Products 
WHERE Size IN ('370ml', '220ml');

A challenge comes up though, when you want to find pairs of values using IN. For example, if I have the following pairs of values:

Size        Color  370ml   Blue 370ml   Red 220ml   Blue

How do I find those when using IN?

Other databases do allow you to have pairs:

SELECT ProductID, ProductName, Color, Size
FROM dbo.Products 
WHERE (Size, Color) 
    IN (('370ml', 'Blue'), ('370ml', 'Red'), ('220ml', 'Blue'));

I wish T-SQL had that option but it doesn’t. It would be particularly useful when those values in the IN clause are coming from a sub-query.

2021-01-26

If you are using a string that was provided by another system, it might not be in the format that you were hoping it would be. So for example, if I have this string that says 2/29/2016, I know that’s a US date format, but if it was only 5/4/2016, there’s no way I could just easily tell that.

Conversion functions use your session settings to determine how to convert the values. When I’m converting strings to dates (and/or times), I certainly don’t want to change my settings so that I run with US configuration.

2021-01-25

I spend a lot of time doing data conversion work. It’s hard to imagine how much of that relates to working with dates and times. A requirement that I had recently was to find the nearest weekday to a given date i.e. find me the closest Thursday to a supplied date.

The NearestWeekday function in our free SDU Tools for developers and DBAs, does this.

It takes two parameters:

@TargetDate date - the date that we’re aiming for @DayOfWeek int - Sunday = 1, Monday = 2, Tuesday = 3, etc. (values from 1 to 7)

2021-01-22

I haven’t posted any book reviews lately but I have been getting through quite a few books, so it’s time to rectify that. First up is Gary Vee’s book #AskGaryVee: One Entrepreneur’s Take on Leadership, Social Media, and Self-Awareness. I often read or listen to books that cover entrepreneurship. If you’ve read any of these, or listened to any related podcasts, you will have heard of Gary Vee. Gary is a Belarusian-American whose name is Gary Vaynerchuk.

2021-01-21

Louis Davidson is an old friend of mine. He’s an awesome Data Platform MVP, a database architect and a prolific author. Louis was on an earlier podcast and I was so pleased to invite him back on the show.

SQL Down Under show 81 features Louis discussing his new book, technical writing, relational DB design, and graph DB applications. Louis’ new book is this huge 1154 page volume:

You’ll find the podcast with Louis (and all our other podcasts) here: SQL Down Under Podcast

2021-01-19

I wish SQL Server had more metadata associated with columns in tables, and with objects in general. The closest thing we have to that is the use of extended properties.

Other Databases

In other databases that I’ve worked with, there are richer properties held for each column. For example in Progress, as well as the normal data type and nullability, there were options like:

• Formatting mask (how this column is normally formatted)
• Prompt (what question to automatically ask the user - gives you a chance to explain what you’re asking)

and much more. This type of thing drastically cut down the amount of time it took to build applications using the database, and provided a great level of consistency.

2021-01-19

In a previous post about CAST and CONVERT, I mentioned that when you use CAST, you don’t get the option to specify the format you want to use. SQL Server offers an extension to the SQL standard with CONVERT and it lets you do just that.

CONVERT has an option parameter that lets you apply a style to the data type conversion that you’re doing.

In the example shown above, I’m converting the current date and time (based upon the SYSDATETIME() function) to a varchar(8), using style 112.

2021-01-18

I do a lot of data conversion work each month. Many systems store date and time values in Unix format, so I often need to convert to/from these in T-SQL. Fortunately it’s easy.

A common way that Unix systems store the date and time value, is to store the number of seconds since the beginning of 1970.

The DateTime2ToUnixTime and UnixTimeToDateTime2 functions in our free SDU Tools for developers and DBAs, does this.

2021-01-15

I’ve attended a lot of user groups over the years. They’ve been data-related, developer-related, and all sorts of other topics. I’ve also organized a lot of user groups over the years. One problem that I fell into early on, and that I see many others falling into, is the problem of increasing depth.

User group leaders often have a very skewed view of the content that’s been presented at their groups. That’s because they’ve often seen almost every single presentation that’s ever been given.

2021-01-14

I’ve seen many challenges with analytics over the years. One that’s painful is an inability to explain where analytic data came from. Someone looks at a report, sees a value, and says I don’t believe that number. Don’t put yourself in that position !

Lineage

I load analytics from data warehouses. Most of my data warehouses are SQL Server databases of some type. Currently, they’re almost always Azure SQL Databases. I like to include information in the database, about how the data got there i.e. the lineage of the data.

2021-01-13