Cocoa (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
momijizukamori) wrote2021-03-23 07:39 pm
momijizukamori) wrote2021-03-23 07:39 pm
Entry tags:
Talking about code instead of working on it
Because all my projects right now are stuck at like.... tedious bits and I'm having a massive moment of 'wehhhhh I don't wanna' so. Gonna catalog them here for now. Feel free to ask questions and whatnot, talking about things may push my excitement about them back up.
Dreamwidth
- Updates to the Select Style/Customize Style pages, with the end goal of better searching/filtering - I have been working on this since like..... 2014, because it devolved into a ton of yak-shaving. The most recent version is awaiting code review, and unfortunately it's big and messy enough that the handful of people who could do it haven't had time.
- REST API - In progress since like, 2017. I have a bunch of stuff like 80% converted, but they need QA and some architectural decisions made, and I have historically been extremely bad at both those things.
- New RTE - the thing I've been devoting most of my brain to lately, but I am at the point where I am 1) hitting up against a bug where the cursor is not going where I want it to and 2) I have to figure out the somewhat complicated logic for exiting block types. Then it will also need more QA, though it's somewhat easier to get that than it is for the API stuff (and it's more tolerant of bugs because it's gonna be released as a beta first)
Lolibrary
- Making it possible to assign items to multiple categories. I know the general shape of how this is going to go, but it involves making some significant DB changes which is.... scary. I might do it this weekend though, I think it's probably a one-day project once I start.
- Better options for search filtering - been kicking around the back of my head for a bit now, but we had a discussion today about a case where being able to exclude based on tag would be really important, so I gotta like... actually write it. Mostly going to be a bunch of fussy query-building.
- Conversion of everything to Python - ongoing project for the last year and a half for me. Currently stalled out around implementing image upload/display. And then, of course, QA.
personal
- Kimono pattern generator in JS - needs an interface for measurement input. Also for me to write up the code for laying out women's cuts and not just the men's. And actual usage instructions, which is the main sticking point.
- AO3 reader for my Kobo - ugggggggh so this is a project that's been niggling at the back of my mind for like a year and a half now. The problem is that I'm working on an embedded Linux system, which means there are limits to the tools available to me. In particular, there isn't easy access to a browser engine for rendering HTML/CSS, which leaves the options as:
1) Use a rendering engine someone else has written and run on the Kobo successfully (there are packages written in Rust and C/Lua, none of which are well-commented, and all of which are languages I don't have much experience in)
2) Write my own rendering engine in Python (iffy performance, and also a huge pain in the ass)
3) Figure out how to cross-compile the correct version of Qt4 and use it's Webkit implementation (this is theoretically the easiest solution except I have spent hours and hours trying to figure out cross-compile and failing)
BUT I really want to easily browse AO3 on my eink reader so the idea won't leave me alone. Very frustrating.
- Fixes to Moody Blues - I need to order a bunch of parts, some of them from a specialty supplier, and with no upcoming cons I just... can't be bothered. I will probably end up frantically trying to understand the code I wrote a year ago in the two weeks before whatever convention I wear it to next.
Dreamwidth
- Updates to the Select Style/Customize Style pages, with the end goal of better searching/filtering - I have been working on this since like..... 2014, because it devolved into a ton of yak-shaving. The most recent version is awaiting code review, and unfortunately it's big and messy enough that the handful of people who could do it haven't had time.
- REST API - In progress since like, 2017. I have a bunch of stuff like 80% converted, but they need QA and some architectural decisions made, and I have historically been extremely bad at both those things.
- New RTE - the thing I've been devoting most of my brain to lately, but I am at the point where I am 1) hitting up against a bug where the cursor is not going where I want it to and 2) I have to figure out the somewhat complicated logic for exiting block types. Then it will also need more QA, though it's somewhat easier to get that than it is for the API stuff (and it's more tolerant of bugs because it's gonna be released as a beta first)
Lolibrary
- Making it possible to assign items to multiple categories. I know the general shape of how this is going to go, but it involves making some significant DB changes which is.... scary. I might do it this weekend though, I think it's probably a one-day project once I start.
- Better options for search filtering - been kicking around the back of my head for a bit now, but we had a discussion today about a case where being able to exclude based on tag would be really important, so I gotta like... actually write it. Mostly going to be a bunch of fussy query-building.
- Conversion of everything to Python - ongoing project for the last year and a half for me. Currently stalled out around implementing image upload/display. And then, of course, QA.
personal
- Kimono pattern generator in JS - needs an interface for measurement input. Also for me to write up the code for laying out women's cuts and not just the men's. And actual usage instructions, which is the main sticking point.
- AO3 reader for my Kobo - ugggggggh so this is a project that's been niggling at the back of my mind for like a year and a half now. The problem is that I'm working on an embedded Linux system, which means there are limits to the tools available to me. In particular, there isn't easy access to a browser engine for rendering HTML/CSS, which leaves the options as:
1) Use a rendering engine someone else has written and run on the Kobo successfully (there are packages written in Rust and C/Lua, none of which are well-commented, and all of which are languages I don't have much experience in)
2) Write my own rendering engine in Python (iffy performance, and also a huge pain in the ass)
3) Figure out how to cross-compile the correct version of Qt4 and use it's Webkit implementation (this is theoretically the easiest solution except I have spent hours and hours trying to figure out cross-compile and failing)
BUT I really want to easily browse AO3 on my eink reader so the idea won't leave me alone. Very frustrating.
- Fixes to Moody Blues - I need to order a bunch of parts, some of them from a specialty supplier, and with no upcoming cons I just... can't be bothered. I will probably end up frantically trying to understand the code I wrote a year ago in the two weeks before whatever convention I wear it to next.
no subject
better searching/filtering to select style sounds super exciting tbh! i’m too old™ to really enjoy building a full style override from scratch in css any more but like, i miss those days.
cursor problems and image uploading problems both make me want to fade into a wall, they’re both things i do not enjoy reminiscing on XD.
i am also over here in the “trying to futilely browse ao3 on kobo” crew (why does it scroll why does it not page), so i will cheer you on.
no subject
Yeah, it's something I've wanted to add for a while, but it turned out adding it involves doing like.... a fuckton of prep work to bring the rest of the interface up to something vaguely modern first, which is why it's taken ages. Also I've chucked like 80% of the rewrite a few times as I get better at programming and realize the way I was doing things was Bad.
The cursor problem is.... fucking arcane. Basically, when you insert a new block type (like a heading or cut or blockquote) at the cursor position, the cursor SHOULD stay at that position and thus be inside the new block. But it's.... not. I posted a question on the ProseMirror discussion board so hopefully someone will have thoughts?
As far as I can tell, the answer is 'because the browser is basically a very thin wrapper over Qt4's webkit implementation, with no optimizations/changes for eink' - unfortunately, nickel (which is the built-in reader app/interface) is the only closed-source bit of the OS, so we can't modify it or see how it's doing the thing. I think Plato is my best bet, because I managed to get it building, and got it to the point of 'load HTML snippet from memory', but it needs to be able to handle links outside the current document, and I don't know enough Rust to do so and the question I posted on their Github issues page..... two months ago, apparently, has not had any responses. So I continue to read fic on my smartphone instead because I don't have enough patience to fight with the browser on eink.
no subject
given that
resolve_locationlets you returnNone, how terrible would it be to open the link from thereno subject
Hmmm I don't think I tried returning nothing from that, though I did try making the request inside there which didn't work for.... some reason (my attempted code is commented out in my local copy, I just don't remember if the error was because I was returning the wrong type or if something else was wrong). Passing the signal to load the new doc up was the other thing that was stumping me, though I may have another look now that I have been getting used to doing crazy stuff with asynchronous state passing with RxJS/ngrx for work.
no subject
icon: tits against the rte Also this has me thinking over ideas again and I wonder if I could make a website that basically proxies requests to AO3 and renders them in a more eink friendly way. Because web apps are the only GUIs I know how to design.
no subject
proxying ao3 might not be the worst idea bc the user interface is not exactly friendly to small screen sizes, though html hates good print-style document formatting 👿
no subject
It kind of depends on how robust the JS/CSS engine is in the Qt4 implementation of Webkit, but! This is the sort of thing I can prototype with a sample fic in a few hours, to see if it's worth continuing down that route, at least.
no subject
no subject
Honestly, I know how to do most of the work for these, it's just.... actually doing it......