25e415daa24a77c4c8a80b11029abeb76f584a45
10
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 prosolis
|
25e415daa2 |
When she writes in Chinese, say Translate — not Clarity
She reaches for her own language mid-sentence when English won't come, and Petal already handled it: it found the span and rendered it into English. It just filed the result as a Clarity fix, so the pair model's flagship moment read as tidying up her Chinese. The type is now derived from the span rather than asked of the model. A type is structural, and a model that re-reasons every pass would drift between labels for a sentence nobody had touched — the instability the last session spent itself removing. The label the model volunteers is still ignored. Only the grammar checkpoint can be promoted. A pass with a forced type owns its family: voice reads paragraphs for tone and its rows carry no replacement, so a "translation" there would be a card offering nothing to accept. zh is a different script and counting Han runes is close to certain. The Latin pairs share an alphabet with English and get none of that, so they fall back to function words and need two before Petal claims anything — with every word that is also English left out, even the common ones. The heuristic is justified by how cheap being wrong is: it changes a coloured pill, and nothing else. The pill is the one bilingual type name in the rail. Every other type stays English because those are the terms she is learning; this card's whole subject is her own language. And it stops truncating its two lines — elsewhere the diff is a word and the explanation is what she reads, but here the two sentences are the card. Two things only the running page could report. The inline underline was invisible: the decoration carries a per-type class and the base rule is a transparent border, so a type with no colour rule gets no mark at all. And at 1517×810 with the document list open there is no rail — the margin is 258 where railEnabled wants 348 — so what she gets is the inline hover card. Item 7 is written the other way round. Migration 0015 rebuilds the suggestions table for the CHECK, which makes it the first one here that could quietly drop her rows; there is a test that carries every column, both timestamps and both indexes across it. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
be1ab5cef7 |
Her apostrophe was cutting French words in half
Typography.ts rewrites every ' typed in the editor into a curly ’, but both word regexes only counted the straight one. So "aujourd’hui" reached the dictionary as "aujourd" + "hui", neither of them a French word, and one of the commonest words in the language came back wearing two red underlines. Same for quelqu’un, presqu’île, prud’homme. l’arbre only survived by accident, because "l" happens to be a bare entry. withElision, written for exactly this, could only ever fire on pasted text. Both marks are word characters now, and combine() straightens on lookup — the one place every lookup passes through — since the shipped word lists spell theirs straight. Suggestions come back wearing whichever mark she actually used, so accepting a pill never swaps her apostrophe. œ was untokenizable too: U+0152/U+0153 sit outside the Latin-1 ranges, so "cœur" split into "c" + "ur" and the orphan was long enough to underline. 586 œ forms ship in fr.dic.gz and not one of them was reachable. In the dictionary builder, the two cross-product paths added their forms without the NEEDAFFIX check the single-affix paths apply, so a doubly-affixed form that is still "not a word on its own" was accepted anyway — the exact class of error the FLAG-aware rewrite exists to close. PFX and SFX are also separate flag namespaces, and one shared `cross` dict let the second block overwrite the first. Odd-length long-flag strings now stop the build instead of dropping a character and expanding through the wrong paradigm. The pt-PT and Québécois greps were case-sensitive against sentence-cased copy, which let a leading "Actualmente…" through the guard added to catch it. Note: this changes what the expander produces, but fr.dic.gz and pt-PT.dic.gz are vendored and were built with the old behaviour. Both want regenerating on a box that can fetch the upstream .deb, and BUILD_PLAN Phase 24's "pt-PT rebuild is byte-identical" claim re-checked — if those bytes move, the NEEDAFFIX gap was live in the Portuguese list too. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
071ea7b835 |
Petal learns French, and the pack that shipped was misspelling itself
Phase 24, the fr half: langpack, Hunspell dictionary, Piper voice, and the lexicon coverage that turned out to have been measured already (63.1%, better than pt-PT's 62.1%). No migration; not deployed. The plan recorded that build_ptpt_dictionary.py "generalizes" to French. It did not. It handled single-character flags and plain PFX/SFX and stopped on everything else, and fr.aff uses four of the things it stopped on. FLAG long is the dangerous one: French flags are two characters, so the old reader's set(flagstr) yields a bag of unrelated letters and expands every entry through the wrong paradigm without ever erroring. Plus continuation flags (French really does affix an affixed form), NEEDAFFIX on 68,075 of 84,140 stems, and FULLSTRIP. Renamed build_hunspell_dictionary.py with a per-language profile, asserting that CIRCUMFIX and FORBIDDENWORD are still unused rather than assuming it — and it rebuilds pt-PT byte-identical to the shipped asset, which is the only thing that makes "generalized" a claim rather than a hope. Elision was decided by building both halves and measuring. Keeping l'arbre and its thirty-three siblings: 3,159,832 forms, 8.25 MB gzipped. Dropping them: 473,326 and 1.19 MB. They are not new words, but the tokenizer keeps internal apostrophes, so they genuinely would have been underlined — so they moved out of the dictionary into withElision, which splits at a known clitic and still requires the remainder to be a word (l'zzzz stays flagged). Real nspell: 369 ms and 74 MB, against pt-PT's 842 ms and 139 MB, on the larger language. Where the regional trap lives is the mirror image of Portuguese's: every fr_* Piper voice is fr_FR and Debian's fr_FR/fr_CA/fr_BE dictionaries are one shared word list, so nothing can be quietly wrong about the country and the whole decision sits in the copy. What French has instead is the 1990 reform, packaged three ways; comprehensive ships, because Petal never corrects her French and coût and cout are both correct. Then the interim review pass, at the user's suggestion and explicitly "for now": four models read each Latin pack independently, and only findings at least two of them reached on their own were applied — five per pack. It earned its keep on the pack that was already live. pt-PT was carrying pre-Acordo spellings (adjectivos, actualmente) in a file whose own header commits to post-Acordo, plus Brazilian decepção, because the Phase 21 greps checked for Brazilian vocabulary and never checked the pack against its own spelling policy. That grep now exists and was confirmed to fail on the old text before being kept. Where reviewers agreed a line was wrong but split on the fix, the wording is mine and the reasoning is in BUILD_PLAN rather than averaged away. Still owed, and both packs now say so precisely: a quorum of models agreeing is agreement, not authority. No native speaker has read either pack, and none of this has been seen in a browser. go build/vet/test clean, tsc, vite build, vitest 190/190. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
1f4ca4775a |
Let her choose her own pair
Raised by the user, not by the plan: there was no way to change language in the mobile UI. There was no way anywhere. `users.pair_lang` has been readable since Phase 19 and writable by nobody — /api/me was GET-only and Upsert deliberately skips the column — which is also why "no pt-PT account exists yet" has stood through two phases. Nothing could create one. PATCH /api/me answers with the whole user rather than 204, so the client re-reads the pair from the server instead of trusting its own request. One write reaches everything: langpack, Hunspell dictionary, Piper voice, lexicon provider and prompt language all read the column at use time. The server refuses a pair it has no copy for, and auth.shippedPairs is deliberately not internal/llm's list. That one names pairs the prompts can talk about (fr and es, since Phase 19); this one names pairs Petal can render itself in, which needs a langpack. Storing fr today would strand her on Chinese with no way back except a lucky guess at a button she cannot read. The picker sits in the sidebar footer because the sidebar is the mobile drawer — always one tap away. The status bar exists only while a document is open, which is the wrong moment to find the app speaking a language you can't read. Each language names itself, 中文 and Português: the one place bilingual copy would get in the way. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
1bbc8fc8d3 |
Finish Phase 22: the half of Petal that works with the tunnel down
Grammar lite, the false-friend list, the daily invitation and the offline miscollocations — the four remaining §5–§6 items, all client-side and all alive on a box that cannot reach the model. The offline collocations forced a schema change. `type` had been doubling as the answer to "which engine found this" — `mechanics` meant offline — and that stops being true the moment an offline rule proposes a collocation. Migration 0013 adds `source` (llm | local) and every pass now scopes its DELETE by engine; without it the coach silently wiped every offline chunk on the page. Existing rows backfill by type, so a pre-0013 collocation row is claimed as the coach's, which it was: the offline list did not exist yet. The rule pack is hand-curated rather than mined, and the entries left out are the point — `married with` is wrong until "married with children", `arrive to` wants at or in depending on the noun. A pack running on every keystroke must not correct correct writing. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
e9b8595456 |
Let the garden keep what she was given, not only what she sought
Two halves of the same idea, both read out of work Petal already records. Planting: an accepted collocation is a learnable chunk, so it becomes a phrase card. The scheduler didn't need to know — a three-word chunk climbs the ladder exactly like a looked-up word. What needed care was deciding what *isn't* a chunk (single words are word choice; a six-word-plus "collocation" is a rewritten sentence, and sentences make miserable flashcards), and that the example must be the *corrected* sentence — the stored draft still holds the phrasing she just left behind. Re-accepting the same chunk leaves the existing card alone rather than resetting a schedule it has been climbing. The whole thing is best-effort: accepting an edit must never fail because a flashcard couldn't be made. The growth journal: kept this month beside kept the month before, the phrasing that stuck, the patterns that faded. The queries were the easy part; the honesty is the feature. "Stuck" needs the phrase in a *second* document, because one document is just the edit where she left it. "Faded" says nothing at all unless she has been writing lately — otherwise a month away from Petal comes back to her as progress, which is the one way this could lie. And a suggestion had to start recording when she *decided* it, not when the model proposed it, so 0012 adds resolved_at and backfills the old rows to their created_at. It lives as a second tab in the garden, and it feeds the kitten: after an accept she now sometimes hears something true of her alone, once per line, half the time, never waited for. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
24c3533e18 |
Give read-aloud a Portuguese voice, and a slower one
Phase 21's infra half. Two things the pt-PT pair needs from TTS, and one thing every learner has wanted since Phase 11. **A language is no longer a code change.** The handler knew exactly two languages, named in the Config struct: English on TTS_ENDPOINT and Chinese on TTS_ENDPOINT_ZH. Petal now discovers its Piper instances from the environment — English keeps the unsuffixed pair it has always had, and every other language is a TTS_ENDPOINT_<LANG>/TTS_VOICE_<LANG> pair — so fr and es cost a compose service and two lines of .env. <LANG> is the base tag, because an environment variable name cannot hold pt-PT's hyphen and only one Portuguese model is loaded either way. A language configured by halves is dropped rather than routed: half a configuration should reach the client as "no voice here, use Web Speech", not as an instance that errors on every tap. The startup line now names the voices it actually resolved rather than the English endpoint it was handed — the same lesson the dictionary line learned last week. **pt_PT-tugão-medium is the only European voice Piper ships.** The other five pt models in the catalogue are Brazilian, so the default anyone reaches for is the wrong country — the same trap as `dictionary-pt` packaging VERO, arriving through the catalogue rather than through the model. Named explicitly in compose, with the query that checks it in the deploy README. **The slow replay** (SUGGESTIONS §5e) is `slow: true` on /api/tts, raising Piper's length_scale to ~4/3. Piper stretches durations rather than resampling, so it stays a voice instead of a groan. The pace is part of the cache key — without it the slow replay of a word already heard at normal speed would be served back at normal speed, which is the one request where the difference is the whole point. 🐢 sits beside 🔊 on the word card, the selection bubble and the garden flashcard; the Web Speech fallback slows too, so the button means the same thing when Piper is down. **And the other reading gets her own voice.** The `alsoIn` block — the Portuguese sense of a word that is also English — now speaks in the pair's locale, which the pack names (`locale`) rather than anything inferring it from the letters. "comum" is spelled identically in both halves; a detector would have to guess, and this is the same reason the gloss shows both directions instead of picking one. Tests: config discovery (both existing deployment shapes, half-configured languages dropped, the pre-map voice defaults preserved), the slow scale and its separate cache entry, pt routing on the base tag with pt-BR landing on the European instance, and speech.ts's request body. The i18n shape suite now asserts every pack names a speakable locale in its own language — and that pt-PT's is not pt-BR. Verified: go build/vet/test, tsc, vitest 125/125, vite build. Live smoke against two fake Piper servers: en/pt × normal/slow all reached the right instance at the right length_scale with four distinct cache entries, and an unconfigured language still 404s. |
||
|
prosolis
|
ccb43e5a4d |
Phase 21: Petal learns to be an English+Portuguese pair
The plan said "Hunspell pt-PT vendored like en-US". Measuring that first is what saved it: nspell expands affixes eagerly on construction, and European Portuguese's 1,340 rules over 44,257 stems want over a gigabyte of browser heap — ~340 MB for the first 12,000 entries, and no return at all after three minutes on the whole file. So the expansion runs once at build time instead: 1,039,058 forms, 2.66 MB gzipped, read by the same nspell in 842 ms. The obvious npm package would also have shipped the wrong language. Both dictionary-pt and dictionary-pt-br carry VERO, the Brazilian word list, so vendoring by name puts pt-BR spellings behind a pt-PT label — the drift SUGGESTIONS §3 warns about, arriving through the packaging where no reviewer can see it. The source is Projecto Natura's, and the build script now asserts the fault lines (receção in, recepção out) before writing anything. Spellcheck consults both dictionaries and flags only what both reject, which is the no-detector answer to a pair with no script boundary. The word card does the same in the other direction: "data" is a word in both languages, so Petal shows both readings rather than guessing which she meant. Writing the tests caught the one real bug — extendedAlphabet was a snapshot while correct/suggest read live, and her dictionary arrives after English, so every lookup would have resolved "cora" while the underlines were already right. Not done, and not claimed: the pack has not been read by a pt-PT speaker, and the Piper voice is deferred with the deploy. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
97e9c269ec |
Phase 20: the dictionary stops being English and Chinese only
Word lookups now come from DreamDict's dict.db for every pair but Chinese — opened read-only beside petal.db, no service, nothing over the VPN, because a hover gloss has to answer in milliseconds. `Provider` is the two questions the popover and the tooltip already asked, so the embedded *Lexicon satisfies it with no changes at all; Set.For(lang) is the single place the choice between them is made. The prerequisite in the dreamdict repo turned out to be two things, not one: the module path was unfetchable *and* the query layer sat in internal/, which no other module may import whatever the module is called. Both fixed upstream. The plan's central assumption did not survive the data. It mapped Gloss ← Translate(word, "en", L1) one-to-one; against the real 452 MB database that table answers for 17% of the 2,000 commonest English words into pt-PT. Wiktionary's translation sections are thin in that direction — "ephemeral", "think" and "quickly" have no en→pt-PT row at all. Shared WordNet synsets answer for 61%, so DreamDict gained Equivalents() and Petal glosses through it. Ordering those was wrong in an instructive way too: sorting by frequency glosses "think" as lembrar, "remember", because lembrar is the commoner Portuguese word even though pensar shares six of think's synsets to lembrar's one. Counting sense agreement first asks the right question. The same measurement is why zh stays on ECDICT: DreamDict reaches a Chinese gloss for 53% of those words, ECDICT for nearly all of them. The plan said converge only if quality holds. It didn't, so nothing converged. Two decisions about failure worth keeping. A missing dict.db is not an error — a laptop checkout has never had one — but a present-and-never-imported one is, because that is a half-finished deploy. And a pt-PT writer with no dictionary falls back to the embedded datasets with the gloss suppressed, keeping definitions, synonyms and phonetics rather than blanking the popover: an empty field reads as "not found", the wrong language reads as broken. The new fields surface as an etymology line and a three-band chip. Three, not five: the difficulty score separates "everyday" from "you'll have to explain this" but cannot rank obfuscate against serendipity, and a finer scale would be a confident-looking lie. An unscored word gets no chip. Writing the tests found two bugs first — trimEtymology sliced by byte, which would have emitted invalid UTF-8 for exactly the Greek and Latin etymologies the feature exists for, and its ellipsis path overran its own cap. go build/vet/test, tsc, vite, vitest 96/96 clean; live smoke against the real dict.db with one instance flipped from zh to pt-PT mid-run. Not deployed: go.mod still replaces github.com/prosolis/dreamdict with ../dreamdict, so the Docker build needs the two upstream commits pushed and the replace dropped. The deployed dict.db also predates DreamDict's Spanish data. Claude-Session: https://claude.ai/code/session_016y6gyuHkQXPiEuW8RGQyua |
||
|
prosolis
|
336cae93e0 |
Phase 19: the copy stops being hardcoded Mandarin
Every `中文 · English` string moves out of ~29 components into web/src/i18n: one Pack type, a verbatim zh pack, and two ways to read it — usePack() for components, pack() for the modules that build a line when something happens rather than when something renders. Anything with a value in it is a function on the pack rather than a template at the call site, English pluralisation included: word order isn't universal, and a pack author has to be able to move the number. The roster constants (tones, rewrite styles, export formats, companions) keep only value + emoji, so a label can't drift from its key. On the server, internal/llm/lang.go replaces "Simplified Chinese" in the three prompts that actually name her language. pt-PT is spelled "European Portuguese (pt-PT, never Brazilian Portuguese)" in the prompt itself, and each Lang carries her word for "why" so the tutor prompt still recognises the question when she asks it her way. pair_lang reaches the model through the row-scoped query each handler already ran — the one that proves she owns the document — rather than a second lookup that could disagree with it. Also records Phase 18's deploy: migration 0011 rehearsed against a copy of the live VPS database, then applied for real. |