Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: Code should be stored in a database. Who has tried this?

news.ycombinator.com

Post reply on HN
Print view View original

21–30 of 51 posts

Re: Ask HN: Code should be stored in a database. Who has tried this?

#21

Post by unilynx »

We did this for a long time for our CMS - although we did simulate a filesystem structure. We also set up a git-like system to store versioning information and set up WebDav to mount it all and allow direct source code editing. It worked pretty well for years.

We eventually stopped because we were relying much more on external tools (eg npm, webpack) which had all sort of issues over webdav mounts. Maintaining all this code management infrastructure in parallel wasn't worth it in the end, and we moved the code back to disk, switched to git, etc.

And photoshop silently ignoring webdav I/O errors when saving designs didn't help either.

You already have tagging by type on the filesystem - the file extension. That allows you to limit file searches. Add extra metadata to extensions if the same extensions have different roles (.backend.ts, .frontend.ts, .html.template, .text.template)

These days I prefer to structure for easy removal of code - everything for eg. a widget (frontend, backend, css) goes into a folder and I only need to remove that folder when the widget is retired, and linting/validation will show me the few remaining path references I need to cleanup.

Re: Ask HN: Code should be stored in a database. Who has tried this?

#22

Post by oliviergg »

827a wrote: post #15

1. The biggest advantage of storing on the filesystem is that Files are the core primitive on UNIX based operating systems, and is core-enough on Windows. Giving that up would require tremendously good reasons. 2. Everyone organizes projects into folders differently, but in most languages the only reason why you organize things into folders is to make it easier for humans to find things. The computer doesn't care whe…

Currently working with salesforce, where code is stored in the database ... to put it simply it's not a really good idea because they never decide to go full db or full files. SO you have to manage files in git and deploy to database. So multiple source of truth : sf db, disk, you git, git of other dev, and central repo. Welcome in 1990 when you work as a team. ie ask your collegue if they already work on component if you need to change it ...

Re: Ask HN: Code should be stored in a database. Who has tried this?

#23

Post by MarceColl »

I implemented something similar for Common Lisp: https://github.com/marcecoll/rekishi

The idea was that you don't have files, just functions that you can bring in and out of scope while editing. You have branches per-function. This all worked more or less transparently to the user using the normal emacs Sly Common Lisp flow.

It was implemented overriding the +DEFUN+ macro, so function re-definitions automatically serialized and created a new entry in the DB.

The Proof-of-Concept used SQLite, but I also envisioned a postgres backed version for jamming on programs with your friends in real time.

Re: Ask HN: Code should be stored in a database. Who has tried this?

#24

Post by Lutger »

You don't have to store things in a database to do this. Code is almost almost read from disk into some kind of in memory data structure that is amenable to such analysis, maybe even more so than a generic database. Doesn't matter if you use vscode or vim, most developers have some kind of tool that does semantic analysis and which affords navigation and organization of code.

Its just that the main way code editors present navigation follows the path hierarchy, also because its often intimately tied to how programming languages shape modules. Most editors have at least some alternative navigation however, and most people are using at least some of them: outlining by declaration symbols, search, changes, unittests, open files, bookmarks, etc.

So in a way, this is already how it is done, except the 'database' part is really tied to the code editor and its storage component nicely decoupled (in the end, databases are usually also just a bunch of files).

I think any real improvements on this model can only come from a new programming language design, and as others have pointed out, this hasn't caught on in the past. The reason for this is probably not that file oriented modularity is the best thing there is, but rather the escape velocity needed to get out of the vast ecosystem of tooling around files, like the OS, git and existing code editors and whatnot.

Re: Ask HN: Code should be stored in a database. Who has tried this?

#25

Post by fergie »

You are talking about taxonomy, and specifically multifaceted classification. In practice a module system with unique names is sufficient- like for example npm.

However its worth noting that all of the systems that rely on databases to store code (SharePoint, SAP, Power Platform) suck haaaaaaard, mainly due to issues with versioning and configuration management.

Re: Ask HN: Code should be stored in a database. Who has tried this?

#27

Post by r24y »

That's basically what an LSP is. It's true that it's built on top of the file system, and most IDE users will navigate using the folder hierarchy, but it still stores information about the name, type, and connectedness of the codebase, and allows querying. Your idea about arbitrary tags (feature, environment) would be useful but does not seem to be supported by the spec [^1] yet.

[^1]: https://microsoft.github.io/language-server-protocol/specifi...

Re: Ask HN: Code should be stored in a database. Who has tried this?

#28

Post by mickael-kerjean »

I have been playing with the exact opposite, representing a database as a file structure where databases show up as top level folders, tables are subfolders, and each row appears as a form like file automatically generated from the schema. You can see a screenshot of such form in [1] which you can edit and save back, effectively enabling anyone familiar with Dropbox to edit data on a database as it just look like a form to fill

The project is oss [2] and the storage connector is "mysql". It even handles foreign key by creating links to another folder with a search query to find the table row it's associated with

[1] https://i.imgur.com/OBJGIeg.png

[2] https://github.com/mickael-kerjean/filestash

Re: Ask HN: Code should be stored in a database. Who has tried this?

#29

Post by AndrewDucker »

In order to uniquely reference one piece of code from another it needs to have a unique name/namespace/reference. Whatever organising principle you use for that will tend to become the hierarchy that your code is stored by.

This doesn't stop you from also accessing it in other ways. And with modern IDEs you can search across a fairly chunky codebase near-instantly, which would allow you to treat it as if it's in a database.

Post reply on HN