Gregory's ServerI had been dreading today's task since the month begun, I knew I'd have to eventually make Drifblim able to survive being thrown paths interactively. The original design expected to be terminated after assembly, relying on self-modification like nobody's business it left the original assembler data in memory nearly unrecognizable after a run. The project here was to make sure that it recovers into a usable state, even after erroring.
|
46 comments
 After completing yesterday's task which was to make it so the little self-hosted assembler was interactive, I knew I'd finally able to build my website without leaving the comfort of Porporo. I did it, felt real nice. But, that's it, I'm calling it a day. ^^ Woke up hungover, booted porporo, slouched with my face resting on my fist, wondering about input methods, if there wasn't an interesting way that I could possibly program with just the one hand holding the tablet pen, that wasn't an on-screen keyboard. I considered the way I could write letters on the palm pilot, and wonder what went into a system such as that, and then implemented it.
[DATA EXPUNGED]
[DATA EXPUNGED]
[DATA EXPUNGED]
@jameschip It's super fun, I think you might appreciate the links I shared on today's Decadv: @neauoire the magic search engine words are “unistroke recogniser” and there’s a whole mess of implementations here https://depts.washington.edu/acelab/proj/dollar/index.html @zens yup! there's a whole bunch of schemes for different setups. I picked one that didn't do any trigonometry.   @neauoire makima!!! One QOL thing if it’s not already implemented, a little backspace button could be a good way to improve tablet-only workflow @slink ^^ I didn't expect shorthand recognition to be such a fascinating rabbit hole to fall into. I've spent a second day on the implementation of such a system. I've wanted to see if the scheme I chose could be extended to cover the rest of the ASCII sheet, and if I could remember each glyph. I've also implemented a punctuation mode. Perhaps I will write tomorrow's entry with a tablet pen. Sir, it's only Day 12, this needs to be stretchier. > Stretch Goal: Build my dream operating system, and have fun! @neauoire I'm also curious to know whether having to use un-natural glyphs like those used for Palm graffiti can be practical for writing code ... Okay, one more day on this, it's just too fun. I've split the canvas into three sections so I can input capitalized letters at the top, lowercase letters in the middle, and punctuation at the bottom. This new UX makes the whole thing a lot more practical to use. I'd like to say that I've written lots of code today, but I mostly practiced drawing the alphabet over and over again. @neauoire Now you could create small drawings for snippet expansion. Imagine drawing something to put the signature in an email or to create and save a file. This month's list of experiments is thinning, one of the remaining thing I'd like to do is create a rom that exposes the controls that have until now only been available through keystrokes to move, close, center, collapse and draw connections between programs. The interesting bit of this project is to have most of the state of the operating system collected into the bar's ram. Since parts of Porporo are built with itself, to compile it we need to have a partial implementation of the Varvara ecosystem capable of running cli roms, and an assembler, so that the roms that live at the core of porporo can be assembled and transformed into valid C files with the help of the formatter rom. When this is ready we can finally compile Porporo. It looks like the work I set out to do on Porporo is done, all I have left is to document it properly. This project is going to transform the way I work moving forward. Today, I completed a handful of challenges from lovebyte's tcc event from within the operating system itself to test things out, without running into any issues. 
[DATA EXPUNGED]
@jameschip it doesn't do much at all, d for drawing lines, m for moving. That's it :) Decided to revisit trying to make one of these 3d tunnels effects after failing last year, with some help from the #lovebytetcc folks, I was able to make a neat 210 bytes demo and learnt a neat trick to approximate distance between two points without a square root function.  max(|x|,|y|) + min(|x|,|y|) ÷ 2 eh? I can see why you like it. It's got kind of a stellar bias, which makes it mysterious. I wish to trade it kelp. 😁 I've done a lot of creative coding these past few days, but today I figured I might do well going over everything with the peephole optimizer, linter, formatter, tidying things up a bit. I've also found a little optimization to the theme loading code that resides at the core of nearly every graphical application I've written.  @neauoire off topic but that is an absolutely gorgeous wallpaper, what is it & who's the artist? @cblgh it's in the alt text of the img tag, but the author is Simon Goinard I guess hand writing and brush strokes haven't quite left my mind since the other day, and so I find myself revisiting Oekaki, I knew that I could use that little trick from monday to approximate distance between points, and connected it to the brush tool to add a quick and dirty natural line thickness. I was helping a friend with their toy scheme compiler last night and we got talking again about SECD, and to help explain the thing, I started putting together a base implementation of the abstract machine's runtime so we could target it with their compiler. It has been a long time since I played with SECD so I've updated the lisp documentation on the XXIIVV wiki with a couple of notes of things I had forgotten. Finished preparing the SECD runtime, I've been wondering if there ever were other functional languages designed to target it besides Pure Lisp(Lispkit), I looked around a bit but I couldn't find any. I noticed there was very little documentation out there on it, so I've started a page of my own. I think I might expand the self-hosted compiler to support line comments and the 'quote expression, because in its current state it's a bit of a pain to use.  @neauoire if you dont mind esolangs, http://www.blue.sky.or.jp/grass/ Wrote very little code today, I closed Henderson's "Application and Implementation", and reopened SICP.. I felt like I needed a refresher of the last chapters(esp. the one about a LISP implementation). So yeah, today was drinking lots of tea, scrolling up and down the wizard pdf, maybe I should get the actual book at some point.. @neauoire ah, i've been reading sicp on and off for the last while. it's a really cute little book. i feel like every time i look at it i gain a deeper appreciation for scheme @neauoire do you have the physical version of it? I wanted to find a second hand copy but it's hard to find or almost as expensive as new @neauoire Just saw it now. It's really interesting, I don't think I've ever seen it before; at a cursory glance it doesn't seem to be equivalent to either manhattan or chebyshev? (Manhattan is dx+dy, Chebyshev is max(dx,dy), and yours seems to be max(dx,dy)+min(dx,dy)/2, if I'm reading it correctly.) I should play around with it a bit and do some comparisons. |
I even went a bit overboard with the recovery pass and I made it so it's optimizing itself a bit each pass. Let's call it .. antifragile software.