Homegrown database? You know how to program so what's the problem?

It's not necessarily a good idea. You'll have full control over something other database creators (whole gangs of them), have a had a hard time creating.

Well, at the time I was looking for a suitable data base, I did not  find any. Either the documentation was incomplete or impossible to understand or, as in most cases, didn't exist. So I made my own.

I wouldn't recommend it.

Now I have a data base for around 38 000 images (and growing!) and my idea was to add some metadata to at least a part of them so I wrote an application for it. It's a data base storing the data as clear text, this makes it possible for anybody to check the contents w/o any tools other that an text editor.

To make it fast all data locations are stored as relative positions in the respective data base file, it's indexed and, yes, it's quite fast. It's also a victim of spurious errors, the window focus may be wrong and you enter something, thinking you're in another window and suddenly that entry is a part of your data base, causing seemingly insane errors.

One day, not long ago I did some changes and afterwards I wanted to change the ownership of some files. I did
"cd <my dir>"
and
"chown -Rv user1:user2 ./*"

The output did definitely not look as I had expected so I did Crtl-c and found that the "." was missing. Not good. That was the last time I logged in to that Linux system.

Here I am, three days later, after reloading my backup, correcting bugs in the program that builds the data base from scratch. I'm closing in on the recovery, slowly and painfully.

It's been interesting, and I had my reasons for doing all this but in retrospect, I wouldn't go down that path once more. All that freedom when designing is just too much for a lonely amateur. Also, being close to 70 years old is not an advantage. I'm slow and error prone.


Sigh!

Update 2020-01-27:
I'm up and running and have fixed some bugs. Once more, I'm an optimist!