And to start of Monday, we have "This database engine does not support fulltext search operations.” when running local tests in sqlite, even though we have verified that fts5 is enabled.
@mjp@nicebyte@solidangle Decided to have some Sunday programming fun with #Vulkan and #sqlite. Took me less than 30 minutes to build a (very simplistic) sqlite based recording and replay system for my samples. I've been working with real time graphics and database for decades, but never really combined them in such a way. Combining those two very different things was a lot of fun :)
I had to implement push notifications on a project that had a particular constraint. Long story short, I can't (can but can't; don't ask) use my backend to store anything. But for this to work, I had to store the subscriptions to the notifications (again, don't ask). And... Enter #SQLite. There's a npm package ready to use and it works flawlessly, so I can have a full fledged, simple and lightweight SQL database right in my frontend. This solved my problem and I'm grateful for this to exist. :)
@kepano Sorry to nitpick but I don't think #2 is valid. If you put enough data in files and directory structures you have a database, though it's a poor one.
All your other points have a clear good/bad angle. With files/databases I don't think it's wise to say files are clearly always superior. If the data gets complex or relational or graph-like, I'll take a #sqlite DB over some mad yaml or JSON thing any day.
You run a #Perl script to generate an #SQLite database with an index of all the #HTML files, publish it, and then embed some #JavaScript which uses sql.js to query the contents of the database and display search results.
@manwar It's called "type affinity" in #SQLite. If you have a column defined as "count INTEGER NOT NULL DEFAULT 0", there's nothing to stop you from inserting the value of "foo" into that column.
It's one of the reasons why I strongly recommend not using SQLite to simulate your database in the test suite: the subtle differences in behavior means your test suite might pass, even if your code is going to fail in production.
We take diffing code by line for granted but every time you have a conflict it becomes obvious that:
it only works because lines are a useful proxy for statements
there would be way less manual resolution of conflicts if the tools would recurse into the offending hunk by token (usefully proxied by whitespace and possibly changes in character class)
As a hobby project to play with #emacs I created a page comparing the words of news stories month to month making a word cloud for each month. I like how it came out but its also a bit depressing.
So each month corpus is compared with its preceeding month dito and word clouds are scaled based on the whole year.
Most of all I developed a new respect for #sqlite which I now think is awesome.
A look at a small section of Curiosity's workspace in this composite image assembled from 4 Bayer reconstructed R-MastCam frames that were assembled in MS-ICE. The images are from December 27, 2023 (sol 4049) at site 105/1108 after a short drive/climb a few sols earlier. The rover is located at the head of Gediz Vallis (see map) I've also attached the drive data. Image credits: NASA/JPL-Caltech/MSSS/USGS/fredk)
@PaulHammond51
I need to redo the database, because I designed it initially to run directly in the browser without a web server, and work directly with NASA's API, but I ended up using a local web server for caching all data from NASA, instead. I expect it to shrink to less than 200MB after I convert more fields to floats.
#SQlite is what I use for almost everything I run locally, even for #QGIS map data. It's a one file db and very easy to work with.
FerretDB enables you to run MongoDB workloads on #Postgres and #SQLite.
Since Neon maintains 100% compatibility with any application that uses the official Postgres release, that makes #Neon an excellent option to consider as the database backend for FerretDB.
Seeing an absurd number of messages in my iCloud backup lead me to investigate the internals of the Apple Messages local database. I documented my attempts to clean up some bad records.
I'm working today on a slightly off-the-wall experiment: can we pack the Crossref database into a usable SQLite file? Working on verifying the schema I've designed against the live data. 75 million records checked, 75 million more to go.
@realn2s
Indeed #SQLCipher and #SQLite "SEE" are not compatible. I learned that from following the former SQLite mailing list and the current forum, but I don't have a link to a specific article.