|
I would suggest a couple of things:
- a stored procedure that is just a collection of these wonderful SQL statements that is commented well and set up to NOT be runnable (comment out the whole thing between /* and */, probably)
- set up a comment area above each piece of code with keywords or details about what the code does and what it would be useful for
- using a search (CTRL-F) on the stored procedure you called GreatCodeToMaybeReuseSomeday (or whatever) to find the SQL of interest.
- Copy and paste the code you found.
Probably the closest to "think it and it will appear"
|
|
|
|
|
Just a thought: Use a library like JsonQL to convert SQL to a JSON representation and store them in MongoDB with descriptive metadata. Search on the content and/or metadata.
|
|
|
|
|
This one gets my "programmer's mind" thinking about a lot of possibilities.
|
|
|
|
|
I find it handy to store them as pass-through queries in an MS Access database. You can give them names that work for you, and then sort them, filter them, search for them, etc. I also wrote a VBA function to let me search for any text string in all my Access queries in case I want to find which ones contain a certain table or field name and so on.
|
|
|
|
|
This sounds like an interesting solution. I figured someone out there must be doing things like this. 
|
|
|
|
|
For very special, or big, or being used a lot, queries (or for clients who can't spell SQL  ), I use something that you will probably hate. I use Java 
I make a program in Java (basically a custom SQL generator) where I insert the code inside a method that receives parameters that allow to configure the query in things like database/table/field names, which fields to return, insert sub-queries, etc. Then document with Javadoc as with any other program.
All the queries are inside the same Java program, are selectable and configurable via command line or a web page that dynamically requests more configuration based on what you requested and configured so far.
Using the generated Javadoc or an IDE on the program source makes it easy to find anything. The program makes easy to get the configured SQL.
This has the advantage that you can tweak the Java source to add more functionality to the queries and make that selectable via parameters. You can even pass the program to clients and they will happily build some queries they need by just answering configuration questions and never touching SQL.
I have done this for a client in the past in which the program would generate SQL and JavaScript (to interact with the SQL via web page) for their web site recommendation system using Similarity Matrices, Friends-of-Friends and a few other algorithms implemented in SQL.
Since their recommendation system used the same algorithms to recommend different things (people, items, lists of people-item pairs, etc), this method was easier for me to maintain and they were very happy to be able to generate the required SQL in less than 5 minutes by answering a few questions instead of manually modifying each time the almost 500 lines of SQL code for the Similarity Matrix alone. They were even able to use the program by themselves, without requesting my help, to generate SQL to recommend things that were never mentioned to me.
Unfortunately, I can not show any of those programs as they are protected by IP 
Yes, this is convoluted, but makes things easier on the long run.
|
|
|
|
|
I don't hate Java, actually. I really like Java as a language. Also, this sounds like the kind of thing that I was thinking might be going on out there -- custom solutions that help you design sql queries and manage them(for devs).
|
|
|
|
|
I meant that you would hate the methodology of writing SQL using another language, not the language itself, since it is more work initially. You have to break the SQL code into pieces to insert variables in between, write the program code and document it.
Most of the times I mention this methodology people look at me as if I just told them to program everything in assembly!
I use Java for this since it is very easy to convert to a servlet and have the interface on a web page, although I prefer to use the command line. My knowledge of web technologies is limited and all my pages end up looking like 
|
|
|
|
|
My company uses SourceGear Vault Standard to store tables, stored procs, functions and queries, but I keep a folder of SQL named with a date and purpose. An example would be '20201201_RestoreDataForCustomer'. The date gives me a context to find them more quickly.
|
|
|
|
|
Rusty Bullet wrote: An example would be '20201201_RestoreDataForCustomer'. The date gives me a context to find them more quickly.
That's one of the tricks I've employed too. but, it also seems to point to a problem that is begging for a management solution. Thanks
|
|
|
|
|
I am always late to the party. HAHAHA. But then I get to read everyone elses comments before I make my own.
What I use is a blend of notepad++ or SSMS and multiple fairly large what I call scratch files. For differing companies and projects I create a scratch SQL project which is mostly commented out sql that includes databases, SPs, Views etc... and then one or 15 differing select statements/kewl sql statements that work for this database scheme.
Then I use Agent Ransack when I am searching for something esoteric that I need for a new client/project.
It works for me and I usually end up cobbling something together from inside the scratch file then copying and pasting into a new SQL for the new SP or View and away I go.
What works for you I guess. PS I do love Agent Ransack.
To err is human to really mess up you need a computer
|
|
|
|
|
rnbergren wrote: It works for me and I usually end up cobbling something together from inside the scratch file then copying and pasting into a new SQL for the new SP or View and away I go.
Yep, that's what it's like. That's why this is interesting to me because it seems like there would be one good solution, but it takes a lot of work to bring together a good way to manage it all. Thanks for your input.
rnbergren wrote: PS I do love Agent Ransack.
another poster mentioned this and i'll be checking it out.
|
|
|
|
|
Save them as .sql files and give them good file names. Then organize them into folders. In other words, treat them like any other application source code.
|
|
|
|
|
That's really the main way to do it. I just don't like sql too much, often forget it and find that since I only have to design queries every few years I am very bored by it.
So, if I could find a way to manage them so I could just find the one I want very easily I would be happy (and it would promote my laziness). 
|
|
|
|
|
Why not store them in a database?
You could use SQL to find the best match...
|
|
|
|
|
Other Dev: What'cha doing?
Me: Writing a SQL query to find a SQL query that I can use in this project. 
|
|
|
|
|
What, can't you just ask the computer "give me this data" and it writes the SQL query for you?
"Hello, computer."
"Use the keyboard? How archaic!"
I hope y'all know what movie I'm quoting. 
|
|
|
|
|
Marc Clifton wrote: can't you just ask the computer "give me this data" and it writes the SQL query for you?
I know. It's quite annoying that I can't. Even Google Home (assistant) won't just tell me answers a lot of the time.
Marc Clifton wrote: I hope y'all know what movie I'm quoting.
I tried DuckDuckGo and got nothing for those quotes.
Maybe, War Games with Matthew Broderick?
Or maybe Star Trek..."Computer...what is our heading?"
|
|
|
|
|
evernote or onenote since they allow for naming sub tabs, I can categorize them, make it a little faster to search
|
|
|
|
|
Matt McGuire wrote: evernote or onenote since they allow for naming sub tabs, I can categorize them, make it a little faster to search
That's an interesting one. I think whatever organization system a dev finds that works for him/her is good. And most people in this thread have said this is basically what they do or just put them all in a text file.
These methods just feel like only one step above keeping them in a plastic binder.
I wish there was a way to really categorize, organize, digitize and systematize the whole situation. But, I'm lazy and slow and I forget SQL all the time (because it is so forgetable).
|
|
|
|
|
Since I'm not the DBA here, I tend to forget how to do simple things like create a table with a index, or how to use cross apply with a function on a table. I only touch these things once in awhile, I needed a way to organize my notes to remember how they work.
It seems once I'm writing SQL it all comes back pretty quick, but when I've been working in code for a few months, all that SQL gets archived in the deep back of my brain, next to the cobwebs and cassette tapes
|
|
|
|
|
Great post!
Your post is exactly what I go through constantly. I may not really need to design and write a new query for 1 or more years. Then I go back and have to invent the wheel all over again.
|
|
|
|
|
I use a private GitHub repo, and use Agent Ransack on my local laptop to find specific files by key words or phrases.
|
|
|
|
|
MSBassSinger wrote: I use a private GitHub repo, and use Agent Ransack on my local laptop to find specific files by key words or phrases.
That's a good combination. Easy to get to from anywhere on the Internet and also you can search them. I'm going to have to look into this agent ransack thing, but the word "ransack" has really put me off it. Many here have mentioned agent ransack.
|
|
|
|
|
Create views with meaningful names?
Isn't that what views are for?
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
