The phrase “SQL programmers” is so fucking weird. SQL isn’t a programming language. It’s a query language. You don’t “program” things with SQL. You utilize SQL as a component of programs for data insertion and lookup, but the actual logic of execution is done in a programming language. Unless you’re doing Oracle PL/SQL, in which case why are you giving money to Oracle?
Edit: Damn, this comment made people mad.
So is Tex. And, yet, I still don’t put it under the “programming languages I know” section on my resume. Probably because it’s not a programming language.
Where you put it is not my problem.
The general census is that latex actually is an example of programming languages sharing semantics with non programming languages and not being intend as a programming language.
since you linked to wikipedia:
The domain of the language is also worth consideration. Markup languages like XML, HTML, or troff, which define structured data, are not usually considered programming languages.[12][13][14] Programming languages may, however, share the syntax with markup languages if a computational semantics is defined. XSLT, for example, is a Turing complete language entirely using XML syntax.[15][16][17] Moreover, LaTeX, which is mostly used for structuring documents, also contains a Turing complete subset.[18][19]
Sometimes even non Turing complete languages are considered a programming language but Turing completeness usually is the criteria agreed upon:
The majority of practical programming languages are Turing complete,[5] and all Turing complete languages can implement the same set of algorithms. ANSI/ISO SQL-92 and Charity are examples of languages that are not Turing complete, yet are often called programming languages.[6][7] However, some authors restrict the term “programming language” to Turing complete languages.[1][8]
You don’t “program” things with SQL
Why not? It sounds like you haven’t written any OLAP queries :)
I’ve written ETL data pipelines using a system similar to Apache Airflow, where most of the logic is in SQL (either Presto or Apache Spark) with small pieces of Python to glue things together. Queries that are thousands of lines long that take ~30 minutes to run and do all sorts of transformations to the data. They run once per day, overnight. I’d definitely call that programming.
Most database systems support stored procedures, which are just like functions - you give them some input and they give you some output and/or perform some side effects.
thousands of lines long that take ~30 minutes
Oh yea!!! Well I have 76 lines of code that takes up to 18 hours to run for 1 client!!!
/s
Your knowledge of data engineering may be limited. SQL is predominant in data processing nowadays. FOSS tools such as DBT allows to write efficient data processing pipelines with SQL and some YAML config without the need for a general purpose coding language.
Why would anyone want that? Because SQL has the interesting property of describing the result you want rather than describing how to compute it. So you can put inside the database, a query engine with decades of optimizations, that will make a much better job at finding the best execution plan than the average developer.
It also means it’s easier to train people for data processing nowadays.
T-SQL is turing complete. While the MS SQL server has limitations on OS level operations, if you allow yourself some leeway with CLR wrappers for the win32 API, there’s no reason I can think of you wouldn’t be able to get the database engine to be a webserver reacting to incoming requests on port 80, or drawing GUIs based off of table state.
It’s be slow and terrible, but doable.
MS SQL Server has this thing called Replication. It’s a feature to keep tables in sync between databases, and even database servers. There’s merge replication (two way), snapshot replication (one way scheduled publishing), and transaction replication (one way live-ish publishing).
And the logic is all implemented in T-SQL stored procedures.
I fucking hate it.
This doesn’t make sense to me. SPs and functions are in every major database. If I wrote a bash script that runs like a program, and sounds like a program, did I program it? Script it?
And lots of systems have nested logic in the DB, optimization often leads to that to reduce overhead. Unless you’re being lazy with an ORM like prisma that can’t even join properly.
Getting high performing queries is just as difficult as any other programming language, and should be treated as such. Even Lemmy’s huge performance increases to .18ish came from big PG optimizations.
Admittedly, this discussion is more one of semantics than anything. It’s pretty clear I’m arguing that SQL is not a “General Purpose Language,” and that proficiency in that domain is what constitutes programming. Which, yeah, is arguably somewhat arbitrary. But my point is that, colloquially, someone who only works with SQL isn’t a programmer. Data Engineer, sure. DBA. Also, sure. Depends on what you do. Programmer? Not really. Not unless you (as in the person, not “it’s theoretically possible”) can use raw SQL to read in video data from a linux system device file and then encode it to mp4 and just nobody’s told me.
Do that in Javascript. Or HTML. Or CSS. Or by that logic is a web developer not a programmer? What about microcontroller programmers?
I could easily write a full logic program in SQL where the API just feeds it data, which is the inverse of how you treat SQL. Admittedly that’s not as common, but it happens pretty frequently in areas of big data, like medical.
I’ve hired Senior Software Engineers that were DBAs, and others that weren’t. They were a development team, all programmers in their own right.
Most database engines support stored procedures. You don’t need to give money to oracle, you can give it to Microsoft instead.
