The reason I want it is because most large videochat solutions have a really high variance in latency and reliability. If I just want a small group chat with friends, I'd have far fewer issues if I just self-hosted. I've already found this to be true for Discord audio versus self-hosting Mumble.
In many cases, I would want to access remote data sources using the SQLite command line, by SQLite extensions which provide them as virtual tables so that SQL queries can then be performed on them including joins with other data sources, copying them to local files, etc (all of this comes "for free" once it is implemented as a virtual table; SQLite does the rest).
Something that can find repetition in a codebase, which could be extracted. Fuzzy matching for duplicated code would also be a great feature too. Not sure if exists, but it's something I've always wanted
I paid for Anki iOS app but the interface is so clunky (even though experts swear by it) that I never really got good use out of it.
I have the desktop app too (on Linux), which I use for deck creation. The interface is really archaic and unattractive (reminds me of an old Tcl/Tk app, though it's written in PyQt). I spent a week trying to get used to it, and I did sort of get the hang of it, but the experience wasn't frictionless enough for me to feel like I want to use it all the time.
I feel that most people who use Anki do it because they are forced to (for med school, for language learning), so they put up with the interface and eventually get used to it.
Anki is kind of a weird local minimum, because there have been many attempts to recreate new "Ankis, but with a better UI" but none have managed to dethrone it.
The closest thing I've found to an "Anki, but with a better UI" is Quizlet. It's not free ($23/year) but so far I'm enjoying the experience.
AnkiDroid is amazing. ~14k cards, so I've committed a lot to Anki.
https://senrigan.io/blog/everything-i-know-strategies-tips-a...
I'm likely not going to use Anki but many of the tips and strategies in your article resonated with me. Thanks!
Box.net have an official OpenAPI descriptor at https://github.com/box/box-openapi , that you can one-time feed into an OpenAPI-to-PHP codegenerator such as https://github.com/janephp/open-api or https://openapi-generator.tech/ to produce a PHP SDK.
This should help address the "relatively unmaintained" part, but the disjointedness is perhaps going to rely heavily on how idiomatic the generated code looks (and you will almost certainly want to ensure there is low-memory/streaming support for file upload/download APIs).
I'd also like an app that I can take a picture of a recipe on paper and then scale all the ingredients by some multiple (half, triple, etc) (obviously also converting ounces/tablespoons/cups/etc).
0: https://github.com/rust-lang/rustlings
1: https://repl.it/github/rust-lang/rustlings
2: https://github.com/rust-lang/rustlings/pull/471
A couple of months ago my friends and I tried to get rustlings working in repl.it . I really, really wish I had documented those efforts thoroughly but at the time it was just an exploratory deep-dive to determine if I should follow up with documentation of issues and more detail, and I decided against using my time for that...kind of regret that now.
Three issues I remember facing:
A) inode not updating in the shell even when files changed in the GUI editor. [seems improved now]
B) /home/runner/.cargo/ (and some other supporting directories i think) disappearing after some time / series of actions and-or inaction [today only via inactivity?]
C) At some point we realized the shell wasn't a 'real' shell but I can't remember what led us to this conclusion, just that it got in our way when debugging the above issues.
(A) [Basically fixed already] What I remember is that the repl.it shell environment, while being extremely high fidelity, seemed to have some idiosyncracies in the way the filesystem worked vs a typical bare-metal OS. Two months ago I had trouble where 'rustlings verify' worked okay but 'rustlings watch' wouldn't notice that a file had been edited. Oddly enough, any shell command investigation would also report that the file hadn't been edited/modified recently, but 'cat file.rs' would still output the saved changes.
(B) [Partially fixed, still an "issue" due to long re-builds but not hard blocker] /home/runner/.cargo disappears after any inactivity. This seems more predictable now than I remember it two months ago - two months ago I might only be able to run 'rustlings watch' once and then very shortly after, the shell would no longer be able to find the rustlings binary anymore and I'd have to rebuild from scratch. Generally there were some errors whenever rebuilding, so the solution for me was typically to create a whole new repl.it from scratch and do the initial build. I don't remember what errors caused the "need" for a new repl.it from scratch every time I needed to rebuild it. This still seems like an "issue", sort of, that the rustlings .replit file has a workaround for which works reasonably well and I suspect this behavior is fairly necessary for repl.it the business. But the binaries still disappear after a short inactive period and need to be rebuilt which takes quite some time on the free tier (10+ min?).
(C) [null, and generally not an "issue"] No memory of what behavior caused us to give the shell the moniker "poverty shell" but overall I've generally found the repl.it shells to be extremely high fidelity such that I generally don't realize I'm using a slightly handicapped system. However, this seems to have changed for python shells...now I seemed locked into something like IPython and can't get to a linux shell via exit(), ctrl-D, etc. Maybe this was always the case and I only remember using the linux shell in the C/C++/Rust repl's.
0: https://github.com/rust-lang/rustlings
Payment method support: stripe and PayPal are enough.
But must have shipment tracking maybe through integration from some service.
All Shopify alternatives suck.