From 4ccf903ff0a82e0c4d3c0f15ceec75f1e7442f6e Mon Sep 17 00:00:00 2001 From: Steve Date: Sat, 10 Feb 2024 18:49:10 +1300 Subject: [PATCH] a 2023 --- G/Lafr/a | 1188 -------------------------------------------------------- G/Lafr/a-2023 | 1189 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 1189 insertions(+), 1188 deletions(-) create mode 100644 G/Lafr/a-2023 diff --git a/G/Lafr/a b/G/Lafr/a index e6d0a65c..62a70e50 100644 --- a/G/Lafr/a +++ b/G/Lafr/a @@ -1,1193 +1,5 @@ -2023 -mo Feb 20 - burning horse afterglow -mo Feb 27 - decide to write a good for a while - get letz up to 45e6ecb2d71d6a - Le is a js object! my first in ages - so that was going explodingly... - in text, everything getting Aip, channel reorganisation bursts - as a ws+lv Linesing Aip,s,c,sc - so letz should parse its own text-sync protocol - projectible clouds of ling particles? - tu - yes, looks good that I stopped - `gitk 64ff01c1d9947...4cd6a3e59f14d` - seeing among unpushed commits noise like: c73267528 - &wind_A 5efe244 and 24e6f21aa - the core of the iterator. - needs to lock sections for progress of others. - ie each one needs to be async and go async for other itselves. - that is the way to avoid having queues, esp such that work repeats. - so the limiting time factor is to allow others to cognise what just happened. - in so far as they have hooked into knowing what happens around them. - then Inside / running / podman begin 35d3a9e6 - and end about 4962811a lezer teeters - a couple of days of letz, then gap. - nearly a month later I commit news of hard drive crash! 5875493c253 - which I basically recovered from to dev letz a little more..? - and lots of holiday not-ing - which would be a great onramp to mi micro teia de registro - time each line, batch by chronohomie - we - cabinny, go to Yours, - th - my office. - resolves to make gondolas pulling film out of piles - to attach labels, put in places, frame - via letz parsing io for thems... - the S prompt is a place! - the screed is the playwriting of the action - it should be a bunch of Somethings - way is stacked up like a micro teia de registro - https://www.youtube.com/watch?v=Rla2CH-1aqI (Eno - Thursday Afternoon) - A watcher is needed... (as above) - and as chatGPT advises - "you will need to figure out how to use the infinitely long IP address to recursively traverse the system" - It seems that "c" is used to define the base identity or characteristics of objects in the program, and it is not clear from the description how these properties are derived or defined. - It seems that "A" objects can contain multiple "C" objects, and that properties can be tallied across them, usually grouping like "C" things. This allows for the expression of many "C" objects on the screen as a stack of individuals bulging around and/or into the items of identity that can be nested in space. This may involve solving a small 2D tiling problem to express efficiently the composition of a bunch of information. - fr - chatGPT - hierarchy can facilitate communication and decision-making by establishing clear lines of authority and responsibility. By creating a clear chain of command, individuals can understand their role within the organization and work together more effectively towards shared goals. - -mo Mar 6 - one year ago I was in to &Ploy - which (felt like a large) space of qua shuffling - was missing io to make all the linkages breakdownable - I still dont have a notey link space - for to get to the T rooms I forget - ie how to play with &Ploy - iirc its a viewer for changed test results - it ends up looking at very samey stuff - we constantly need to compress the output - as soon as we have some pattern over the chaos - the frequency wants to rise again, thussing and thussing - I was making time-socket S|Go - showing what haps with -AXV - talk to chatgpt https://howsecologysquad.blogspot.com/2023/03/cgpt-roundup.html - in conclusion I would say: - voicing() seems to be my goal - some apparatus of multitudinous movementisms for the mind <-> world (the world of A+C). - perhaps it needs to be molecular, a few A walk each other seeing what to do to the other. - to split the 'thinking of you' - as in, this was a waste of a week. oopsies. -mo Mar 13 - so want to go to Karamea this weekend. - I come along sometimes and the computer is going oom - transmission 3.00 has 15.7% of it - `free -h` says available 233Mi, which is low. - tu - I read node_modules/@codemirror/commands/dist/index.d.ts - interface MarkDecorationSpec { - are for wrapping text in multiple - interface WidgetDecorationSpec { - for putting in text - interface ReplaceDecorationSpec { - for drawing closed chunks... - we will need to transmit ghostlets with closed chunks in. - perhaps metadata describes where and how many lines are missing - the included lines should address the opening - this would look just like spec expanding an inner branch from some topic - just like this line here, which neednt be except to say this more clearly - declare abstract class WidgetType { - all good. - https://codemirror.net/docs/ref/#state.EditorState.update - (...specs) Create a transaction that updates this state. - https://codemirror.net/docs/ref/#state.EditorState.phrase - shortcuts with params - https://svelte.dev/docs#run-time-svelte-setcontext - ~~ ark, localising Names, &t lookaround. - is &t for u or n? - Context [not] reactive [...] you can pass a store into context, which will be reactive. - seems like large items (G) could sit in the codemirror where they are configured - https://learn.svelte.dev/tutorial/page-data - Every page of your app can declare a load function - in a +page.server.js: - export function load() { - return { - summaries: posts.map(... { slug..., title...}) - - the +page.svelte can access this via the data prop: - -
    - {#each data.summaries as { slug, title }} - - then the one src/routes/blog/[slug]/+page.server.js: - import { error } from '@sveltejs/kit'; - export function load({ params }) { - post = findpost(params.slug) - if (!post) throw error(404); - return { post } - - and the one src/routes/blog/[slug]/+page.svelte: - - -

    blog post

    -

    {data.post.title}

    -
    {@html data.post.content}
    - - Every one of these Svelte docs pages has a worker, it looks like. 400Mb - I write https://howsecologysquad.blogspot.com/2023/03/bc.html in the evening - we - GPT4 release - 32k tokens now, will surely help my rabbit-warren light up - a non-event: not clear how I can use it today - chatgtp attitude - have languished in dialing chatgtp - need automation - more https://learn.svelte.dev/tutorial/the-form-element - great simplicity - you (as letz/src/routes/sverdle/+page.server.ts): - export const actions: Actions = { - update: async ({ request, cookies }) => { - or s/update/default/ and drop action from
    - even: - does great paving over the jagged whoirley of http+html this used to be - so the form empties when submit... - to keep values in, have them come back from the server in: - - supposedly this works with javascript off? - and render - I go to The Dog, the dying open-mic-having cafebarvenue in the evening. - it was a bit griefy and I part with a Rye from Yours, seeking beans over people. - th - sa - Svelte slots seem good. - can I read the objects out of them after html layout but before render - and rearrange things until I like it? - su - remembering yesterday, - the voice looks back in time until it recognises something, calls that parent -mo Mar 20 - Svelting up - @letz b6b812dcee more animation (using) - https://youtu.be/3TVy6GdtNuQ?t=2054 - passing events around takes lots of labels - surely theres automation for this slinking of fairylights - tu - https://dagshub.com/ is looking good - see src/WebGL-Fluid-Simulation bb4b6d8009fccd -mo Mar 27 - buth day! - tu - @letz: -mo Apr 3 - th - o_up() - we Apr 12 - cont. - fr -mo Apr 17 - @letz / St.ts - make class A|C to facilitate their id-ing - it seems to be free with an object - tu - @letz / Con data dumper - we - being ready to Conbustank - th - a little Conbustank - @letz / Con - ping the innards! my datadump is not re-rendering - su -mo Apr 24 - {#key} at the top level (redo everything) is all that works!? - tu - we will have to do lots more of Con in St... - we - exporting a C** just for Con - th - unifying C** interface - fr - solved dispatch problem - wrote up blog on this week - https://howsecologysquad.blogspot.com/2023/04/svelte.html -mo May 1 - Conbustank fra- - tu - Conbustank -ming - we - @letz / Con - still not working - th - the Cont%Ct is stale, its y&up.c.s.t is not... - figure out a bunch more of my faults: - we need to send the new -Con - with a .c.version - < that is real - reactive the o -Con/* - su - thoroughly repaired yesterday - @letz / test.ts - towards `npm run test` - deps: npx playwright install; npx playwright install-deps - < test build having that done - mo - test passing - real 0m37.355s real slow. how to hold the loop? - tu - @letz 036941112b / St<->Con - class sip_dispatch - has fixed "stop wires from being recycled" - no idea why. perhaps toplevel is recreating? - critical to updates is: Code.svelte: {#key conver} - after a second bleep() no updates received - perhaps the old con is still used..? - so is Con.svelte: {#key quee} - which basically dispatches to Con/Cont etc for us - we - @letz / St<->Con - should check we can dispatch updates to sip=1 - and lo, problems with 1.2.5 (the new dat.i) not having a wire - ie the throw unwired in Con.svelte - the wire has been made but unfound by getContext - creating a new Con//Con can cause old Context reuse apparently.. - allowing this to happen now reveals: - always unwired after refresh (and bleep, bloop) - further bloops keep it unwired - another bleep, bloop+ always works - kind of stabilise updates to sip=1 @letz 5e54e1dbdbb7 - ensure refresh=0 during bleep() - we were able to ignore the 1.2.5 spawn - during loops of (bleep,bloop)+ - since 'refresh' would stay at 1 - thinking about diffing... - be nice to see Con output in Con - wondering what properties they have... - just each other, their t,cv,sc - d.visit needs to include (and will only use) -Con[tz] - d.D is the d.C from before - last output fed in again at the top - d.D must 'resolve $n' in time to find -Conz its D.c.next_boost - C.c.X as the absorber of everything else on C.c|sc.* - going into those things as far as it can, eg sip - avoiding other C? - > looking for time-space deals in there. is it "the same"? - Con sees little - Cont has the label, typology - Conz has a some children - and must template any rearrangement? - X.c|sc=$gk - .is=v if nonref - .sym=such,k=$k - could then turn into an {is} or {sym,k} if more hash... - .sym=such,k=$k2 - .is='blub' - eg is C.sc.flab.$k.$k2 = 'blub' - ignoring C&Cont etc - C&X <=> D&X - th - keep it simple now... - bind:this might help make an index of components - which we might simply call for an update on... - instead of this wire stuff - fr - @letz 358b75b82f9b / Co.ts - do DCresolve and DCdiffer - wire all -Con** and remove their {#key} - sa - 763d2f4b7 fixed: deleting mind (1.2.2) keeps its label, - seems to misalign Cont<->Conz for several things - probably need to check D<=>C ip somehow - we just needed to put Con.svelte C.* reading in a $: upto(C) - {#each} can have a key, items in a changing list remain themselves - goners going, next item slides up - looking at it do transitions we are mutating the goner -mo May 15 - we - screen record some demos illustrating last week - @letz - 11c6bb1c272 remove goners and rename it sip_dispatch.setN - fixed: old sip is updating as it fades out - its size|content becomes that other - and doesnt get removed after transition (usually) - interestingly, noticed this only becuase: - 78b26a563ec cause lots of el:4 spam to every ~C.c.ip - which is obsolete, since 86f3f76209d5 - C.c.ip can be kept in sync without dispatch if put in upto() - svelte seems to be running all these upto() functions every time - luckily they are just assigning a few strings... - < get definite svelte<->data awareness awareness - subscription dupe from the C now assigned its ip - would otherwise be more cornery - < fuzz testing - perf - toCon() 1.6ms - setN() 2.2ms - it is all console.log - layout etc 60ms - then without transition 10ms - conver++ (keyed, recreates everything) 80ms scripting, same-ish layout - dramatic revelation: sip dispatch not required! - every upto() is running (anbup) - an intelligent minimum of work (mozwales) - even if the C remains the same (no tocon()) - ie not '==', perhaps because object - < should we make a tree of versions we can recurse efficiently? - I hope https://joyofcode.xyz/svelte-state-management - will answers? no. why doesnt it not work? - th - @letz c435d04b4a - has no need for all this dispatch management - hundreds of things can keep up to date speedily - I suppose the alignment of values in upto() is key... - perhaps all this dispatch will be useful somehow - we certainly are near lv emission there... - now we wander back to what to do... - leaving the dispatch stuff around, disabled - < use it to navigate a tree of versions - - throwing a Le (an AST fragment from CodeMirror) into the Con - boost redo bunch of Con** via toCon_reentry() - that is some A business how it replaces subnets - w,e,r are the keys - bleep() on startup (Mount) - fr - see Inside / and so... / mediaserver detour - sa - as above - su - as above, - letz/py - this works quite well after only a day - makes ~5kb webp thumbnails, takes <1s for videos - letz split Diring from Code - wrong: - in Code.svelte: - import { Diring } from "$lib/Diring.svelte" - vite will say: - Error: is not a valid SSR component. ... - right: - import Diring from "$lib/Diring.svelte" - also if: - {#each peek as f, i} - at the end of Diring causes the same "not a valid SSR component" - along with an earlier: /src/lib/Diring.svelte:69:7 'peek' is not defined - you have to be careful - to not click away the modal error message in the webapp, - whence it will have failed to reload - detour for exploring video thumbnails - backend work in py/serve.py - letz py thumber - 2 in parallel, when timestamp > video, snap at 1s - opencv used simply to take 9 samples - < finding representative thumbs - mo - detour for exploring video thumbnails - tu - @letz / Co.ts / inlace Con frontend work - prototype inlace refactored for tail calls - surprise: no tail call optimisation in javascript - so bloop() now runs in 65ms, not 2.5ms - it gets 172 calls deep for our 20 d - nice picture of the resolving stagger of Atime - ie A1@3 with A11@1 - default d.tailcalls=0 - bloop() now runs in 1.5-2.5ms - < how to depth-first (or not) - will generalise to io expr demanding perspective - with even @2/@3 possible (child runs ahead) - - another inlace frontend for -Diring//-Dir - api wrangler - we - Conbustank painty - @letz / Dir - climb the s* to find all -Dir above us we are relative to - to formlink to the whole thing - th - Conbustank mounty - fr - Conbustank joiny - mo - @letz Compile.ts stylehouse_lite(source) - we seem to already have npm:magic-string - https://inspirnathan.com/posts/27-how-to-use-the-magic-string-library/ - Babz translated: - '#' comments - as a vite thing - @letz - Dir/Img - how do we know its an Img? - we seem to already have npm:mime to suppose .jpg$ - tu - Conbustank paint - we - Conbustank paint - th - just lounge - fr - food mission - sa - Waitati hillside - @letz Compile.ts <-> vite? - su - this seems to be it: - https://github.com/sveltejs/svelte-preprocess/blob/main/docs/preprocessing.md#custom-preprocessors - but the svelte() there is sveltekit() here (rollup ~~ vite?) - it seems to be for svelte.config.js, which has a preprocess - and now we seem to be outputting js - by the way its erroring about typescript syntax - how to have it both ways? lang="ts+stlli"? - its not so obvious that it would become javascript... -mo June 5 - as vite plugin with caveats - vm networking nightmarelet - Im teetering along figuring out the compile-to-javascript - when I lose the ability to connect to http://192.168.122.92:3000/?ierorag=9 - I check chmod u+s /usr/lib/qemu/qemu-bridge-helper - I rebuild cos as user-mode podman, as the py/ is - in src/letz/ is some root-owned stuff to delete: - .svelte-kit - "You can ignore its contents, and delete them at any time (they will be regenerated when you next dev or build )." - also node_modules/ (it lsyncs back again, owned by s) - I also drop node_modules and `npm i` them back - then rsync these back to sa:src/letz/ so I can see into libraries in vscode - this might have helped... - it seems I must use http://[::ffff:c0a8:7a5c]:8000/ now (related?) - it can take a while to realise your frontend (browser at page) is not connected - once you are used to it updating without you hitting F5 - tu - as svelte preprocess, esbuild - with sourcemaps - we - errands all day - kind of conclude - https://howsecologysquad.blogspot.com/2023/05/svelte-further-than-ever-before.html - we have .ts and .svelte +
      + {#each data.summaries as { slug, title }} + + then the one src/routes/blog/[slug]/+page.server.js: + import { error } from '@sveltejs/kit'; + export function load({ params }) { + post = findpost(params.slug) + if (!post) throw error(404); + return { post } + + and the one src/routes/blog/[slug]/+page.svelte: + + +

      blog post

      +

      {data.post.title}

      +
      {@html data.post.content}
      + + Every one of these Svelte docs pages has a worker, it looks like. 400Mb + I write https://howsecologysquad.blogspot.com/2023/03/bc.html in the evening + we + GPT4 release + 32k tokens now, will surely help my rabbit-warren light up + a non-event: not clear how I can use it today + chatgtp attitude + have languished in dialing chatgtp + need automation + more https://learn.svelte.dev/tutorial/the-form-element + great simplicity + you (as letz/src/routes/sverdle/+page.server.ts): + export const actions: Actions = { + update: async ({ request, cookies }) => { + or s/update/default/ and drop action from + even: + does great paving over the jagged whoirley of http+html this used to be + so the form empties when submit... + to keep values in, have them come back from the server in: + + supposedly this works with javascript off? + and render + I go to The Dog, the dying open-mic-having cafebarvenue in the evening. + it was a bit griefy and I part with a Rye from Yours, seeking beans over people. + th + sa + Svelte slots seem good. + can I read the objects out of them after html layout but before render + and rearrange things until I like it? + su + remembering yesterday, + the voice looks back in time until it recognises something, calls that parent +mo Mar 20 + Svelting up + @letz b6b812dcee more animation (using) + https://youtu.be/3TVy6GdtNuQ?t=2054 + passing events around takes lots of labels + surely theres automation for this slinking of fairylights + tu + https://dagshub.com/ is looking good + see src/WebGL-Fluid-Simulation bb4b6d8009fccd +mo Mar 27 + buth day! + tu + @letz: +mo Apr 3 + th + o_up() + we Apr 12 + cont. + fr +mo Apr 17 + @letz / St.ts + make class A|C to facilitate their id-ing + it seems to be free with an object + tu + @letz / Con data dumper + we + being ready to Conbustank + th + a little Conbustank + @letz / Con + ping the innards! my datadump is not re-rendering + su +mo Apr 24 + {#key} at the top level (redo everything) is all that works!? + tu + we will have to do lots more of Con in St... + we + exporting a C** just for Con + th + unifying C** interface + fr + solved dispatch problem + wrote up blog on this week + https://howsecologysquad.blogspot.com/2023/04/svelte.html +mo May 1 + Conbustank fra- + tu + Conbustank -ming + we + @letz / Con + still not working + th + the Cont%Ct is stale, its y&up.c.s.t is not... + figure out a bunch more of my faults: + we need to send the new -Con + with a .c.version + < that is real + reactive the o -Con/* + su + thoroughly repaired yesterday + @letz / test.ts + towards `npm run test` + deps: npx playwright install; npx playwright install-deps + < test build having that done + mo + test passing + real 0m37.355s real slow. how to hold the loop? + tu + @letz 036941112b / St<->Con + class sip_dispatch + has fixed "stop wires from being recycled" + no idea why. perhaps toplevel is recreating? + critical to updates is: Code.svelte: {#key conver} + after a second bleep() no updates received + perhaps the old con is still used..? + so is Con.svelte: {#key quee} + which basically dispatches to Con/Cont etc for us + we + @letz / St<->Con + should check we can dispatch updates to sip=1 + and lo, problems with 1.2.5 (the new dat.i) not having a wire + ie the throw unwired in Con.svelte + the wire has been made but unfound by getContext + creating a new Con//Con can cause old Context reuse apparently.. + allowing this to happen now reveals: + always unwired after refresh (and bleep, bloop) + further bloops keep it unwired + another bleep, bloop+ always works + kind of stabilise updates to sip=1 @letz 5e54e1dbdbb7 + ensure refresh=0 during bleep() + we were able to ignore the 1.2.5 spawn + during loops of (bleep,bloop)+ + since 'refresh' would stay at 1 + thinking about diffing... + be nice to see Con output in Con + wondering what properties they have... + just each other, their t,cv,sc + d.visit needs to include (and will only use) -Con[tz] + d.D is the d.C from before + last output fed in again at the top + d.D must 'resolve $n' in time to find -Conz its D.c.next_boost + C.c.X as the absorber of everything else on C.c|sc.* + going into those things as far as it can, eg sip + avoiding other C? + > looking for time-space deals in there. is it "the same"? + Con sees little + Cont has the label, typology + Conz has a some children + and must template any rearrangement? + X.c|sc=$gk + .is=v if nonref + .sym=such,k=$k + could then turn into an {is} or {sym,k} if more hash... + .sym=such,k=$k2 + .is='blub' + eg is C.sc.flab.$k.$k2 = 'blub' + ignoring C&Cont etc + C&X <=> D&X + th + keep it simple now... + bind:this might help make an index of components + which we might simply call for an update on... + instead of this wire stuff + fr + @letz 358b75b82f9b / Co.ts + do DCresolve and DCdiffer + wire all -Con** and remove their {#key} + sa + 763d2f4b7 fixed: deleting mind (1.2.2) keeps its label, + seems to misalign Cont<->Conz for several things + probably need to check D<=>C ip somehow + we just needed to put Con.svelte C.* reading in a $: upto(C) + {#each} can have a key, items in a changing list remain themselves + goners going, next item slides up + looking at it do transitions we are mutating the goner +mo May 15 + we + screen record some demos illustrating last week + @letz + 11c6bb1c272 remove goners and rename it sip_dispatch.setN + fixed: old sip is updating as it fades out + its size|content becomes that other + and doesnt get removed after transition (usually) + interestingly, noticed this only becuase: + 78b26a563ec cause lots of el:4 spam to every ~C.c.ip + which is obsolete, since 86f3f76209d5 + C.c.ip can be kept in sync without dispatch if put in upto() + svelte seems to be running all these upto() functions every time + luckily they are just assigning a few strings... + < get definite svelte<->data awareness awareness + subscription dupe from the C now assigned its ip + would otherwise be more cornery + < fuzz testing + perf + toCon() 1.6ms + setN() 2.2ms + it is all console.log + layout etc 60ms + then without transition 10ms + conver++ (keyed, recreates everything) 80ms scripting, same-ish layout + dramatic revelation: sip dispatch not required! + every upto() is running (anbup) + an intelligent minimum of work (mozwales) + even if the C remains the same (no tocon()) + ie not '==', perhaps because object + < should we make a tree of versions we can recurse efficiently? + I hope https://joyofcode.xyz/svelte-state-management + will answers? no. why doesnt it not work? + th + @letz c435d04b4a + has no need for all this dispatch management + hundreds of things can keep up to date speedily + I suppose the alignment of values in upto() is key... + perhaps all this dispatch will be useful somehow + we certainly are near lv emission there... + now we wander back to what to do... + leaving the dispatch stuff around, disabled + < use it to navigate a tree of versions + + throwing a Le (an AST fragment from CodeMirror) into the Con + boost redo bunch of Con** via toCon_reentry() + that is some A business how it replaces subnets + w,e,r are the keys + bleep() on startup (Mount) + fr + see Inside / and so... / mediaserver detour + sa + as above + su + as above, + letz/py + this works quite well after only a day + makes ~5kb webp thumbnails, takes <1s for videos + letz split Diring from Code + wrong: + in Code.svelte: + import { Diring } from "$lib/Diring.svelte" + vite will say: + Error: is not a valid SSR component. ... + right: + import Diring from "$lib/Diring.svelte" + also if: + {#each peek as f, i} + at the end of Diring causes the same "not a valid SSR component" + along with an earlier: /src/lib/Diring.svelte:69:7 'peek' is not defined + you have to be careful + to not click away the modal error message in the webapp, + whence it will have failed to reload + detour for exploring video thumbnails + backend work in py/serve.py + letz py thumber + 2 in parallel, when timestamp > video, snap at 1s + opencv used simply to take 9 samples + < finding representative thumbs + mo + detour for exploring video thumbnails + tu + @letz / Co.ts / inlace Con frontend work + prototype inlace refactored for tail calls + surprise: no tail call optimisation in javascript + so bloop() now runs in 65ms, not 2.5ms + it gets 172 calls deep for our 20 d + nice picture of the resolving stagger of Atime + ie A1@3 with A11@1 + default d.tailcalls=0 + bloop() now runs in 1.5-2.5ms + < how to depth-first (or not) + will generalise to io expr demanding perspective + with even @2/@3 possible (child runs ahead) + + another inlace frontend for -Diring//-Dir + api wrangler + we + Conbustank painty + @letz / Dir + climb the s* to find all -Dir above us we are relative to + to formlink to the whole thing + th + Conbustank mounty + fr + Conbustank joiny + mo + @letz Compile.ts stylehouse_lite(source) + we seem to already have npm:magic-string + https://inspirnathan.com/posts/27-how-to-use-the-magic-string-library/ + Babz translated: + '#' comments + as a vite thing + @letz + Dir/Img + how do we know its an Img? + we seem to already have npm:mime to suppose .jpg$ + tu + Conbustank paint + we + Conbustank paint + th + just lounge + fr + food mission + sa + Waitati hillside + @letz Compile.ts <-> vite? + su + this seems to be it: + https://github.com/sveltejs/svelte-preprocess/blob/main/docs/preprocessing.md#custom-preprocessors + but the svelte() there is sveltekit() here (rollup ~~ vite?) + it seems to be for svelte.config.js, which has a preprocess + and now we seem to be outputting js + by the way its erroring about typescript syntax + how to have it both ways? lang="ts+stlli"? + its not so obvious that it would become javascript... +mo June 5 + as vite plugin with caveats + vm networking nightmarelet + Im teetering along figuring out the compile-to-javascript + when I lose the ability to connect to http://192.168.122.92:3000/?ierorag=9 + I check chmod u+s /usr/lib/qemu/qemu-bridge-helper + I rebuild cos as user-mode podman, as the py/ is + in src/letz/ is some root-owned stuff to delete: + .svelte-kit + "You can ignore its contents, and delete them at any time (they will be regenerated when you next dev or build )." + also node_modules/ (it lsyncs back again, owned by s) + I also drop node_modules and `npm i` them back + then rsync these back to sa:src/letz/ so I can see into libraries in vscode + this might have helped... + it seems I must use http://[::ffff:c0a8:7a5c]:8000/ now (related?) + it can take a while to realise your frontend (browser at page) is not connected + once you are used to it updating without you hitting F5 + tu + as svelte preprocess, esbuild + with sourcemaps + we + errands all day + kind of conclude + https://howsecologysquad.blogspot.com/2023/05/svelte-further-than-ever-before.html + we have .ts and .svelte