# @ibgib/blank-gib changelog _note: as you implement features/fixes/etc., please document them here under the "Working Version" section. These will be moved to a concrete version number during the next publish._ ## Working Version ## 0.15.13-14 * refactor: blank-gib bootstrap refactored * implemented `script.mts` for early UI Shell initialization, improving perceived performance. * modernized `index.mts` to use the new multi-phase bootstrap orchestration. * broke early component loading chain in `helpers.web.mts` by removing premature component imports. * updated `index.html` to include the new `script.mjs` entry point. * verified build pipeline with new `script.mts` entry point. * chore: centralized build orchestration * migrated build, clean, and test logic to the root `@ibgib/build-gib` orchestrator. * refactored `web1-links-component` to use native rendering instead of an iframe. * implemented premium styling for links list using root CSS variables. * updated curated links with motivation, NPM (reverse order), and social sections. * added X account (`https://x.com/ibgibDOTcom`) to social links. * implemented dynamic link rendering with "copy to clipboard" functionality. * added YouTube embedded player support for video links in the curated list. * infra: containerized deployment architecture * implemented `Dockerfile` and optimized `nginx.conf` for serving .mjs modules. * configured Traefik labels for automatic routing and SSL termination. * pruned legacy `package.json` scripts and archived them in `docs/ARCHIVE_SCRIPTS.md`. * updated `prepare:publish` to use the centralized build engine. * documented the new **Monorepo Build Policy** in the README. ## 0.15.12 * progress: keystone respecs passing Suite A - D * still working on low level improvements * stopping now at the point before breaking out validate code to also include validating individual frame and validating the entire keystone graph/timeline intrinsically. ## v0.15.10 * progress: testing for keystone mechanics * innerspace and metaspace_innerspace still not functional for testing * I think it has something to do with needing singletons for space ibgibs, which is a non-trivial problem (for any singleton ibgib) * using a mock metaspace and space, seems to be working okay * Suite A and B tests are passing * Currently working on Suite C which is failing. We are moving to a better validation function that will provide validation errors instead of an opaque boolean. ## v0.15.6 * impl: upgraded build to handle respec-gib testing * npm run test did not work, as this was still configured for the simpler pre-esbuild build process. * added build-test.mts as esbuild target and tsconfig * progress: keystone plumbing * restarting bottom up keystone with Gemini 3.0 collab * https://aistudio.google.com/prompts/1-Itqn0afYpgnm-ajEID8Mv0SAHii7OBY * largely me driving higher level and getting Gemini 3.0 up-to-speed on understanding the architecture enabled by the ibgib protocol, with Gemini iterating on implementation details. * the factory style approach was Gemini's idea, which is a good tactical decision IMO * deferring refactoring most names until after, as that causes hiccups with LLMs * doing in blank-gib/keystone for now for ease of development * this was the driver to get the testing working for blank-gib using the new esbuild process * meta: iterating styleguide * using entire styleguide as part of system text for Gemini in Google's AI Studio ## v0.15.3 * **Refactor: Rabbit Hole Comment Architecture** * Changed details views to toggle-based. Now you can see tldr AND translation AND key points. * Overhauled the component to use a consistent, template-based system for all detail views (TLDR, Translate, Key Points). * Views are now created dynamically on-demand, improving initial render performance. * **Fix: collapsed `ul`/`ol` tags** * Short list tags were getting collapsed and losing navigation. ## v0.15.2 * **Feature: In-App Translation** * Integrated the built-in Chrome `Translator` and `LanguageDetector` APIs. * Added a "Translate" button and language selection dropdown to each comment component. * The source language is now auto-detected and hidden from the target language list. * Translations are performed on-device and saved to the comment's `ibGib` data for persistence. * **Refactor: Component Logic** * Refactored the `switchToView` method into smaller, more manageable sub-methods (`switchToView_children`, `switchToView_summary`, `switchToView_translation`). * Created new helper functions (`getTranslationTextKeyForIbGib`) to handle complex data key generation, preventing parsing issues with underscores. ## v0.15.1 * working on extension for hackathon built-in AI from google * changed generate-version-file.js from using `assert` to `with` in import statement. * this was due to a node build warning saying that `assert` was being deprecated. ## v0.14.1 * refactored to esbuild * had to tweak component architecture a bit * no more dynamic JIT loading of css and html, now these are AOT * esbuild config to not mangle names (`keepNames: true`) ## v0.13.1 * backing up before trying to migrate to esbuild * for perf reasons * also to get the app to work in chrome web extensions, which don't currently support import maps. * not sure the other changes before this version, but I believe they were minor adjustments and I was publishing them to ibgib.com (thus the version increments). ## v0.8.1 * parrot, fill-in-the-blank (fitb), and translate added to stimulus and gameplay * all extremely rough, and it's been slooooooow going. * moving variant from the game itself to each of the stimuli * hard-instantiation of fitb entries right now, but need to do dynamic ones * only superficially tested, but it's been so long I'm throwing it up there and going to test it out on my own real-world data. * this reminds me, I don't truly have a migrate functionality, so we'll see how that goes eessh ## v0.7.1 * simple metronome implemented * display version on bootstrap ## v0.6.1 * working on divergent agents among other bugs * deprecated functions are now just checked at runtime instead of mutated within the agent * until the agent divergence issue is resolved, then we can go back to pruning them "properly". ## v0.4.1 * chat active tab and persist open tabs per project in chronologys panel are working now. * had to correct some of the shared plumbing for ParentOfTabs components. * still need to migrate over the projects to include closing tabs, but this is good enough for now. * next up is focusing on manually creating a pipeline for learning source text. ## v0.3.1 * chat active tab * persist open tabs per project in chronologys panel * still buggy but hey, there ya go ## v0.2.1 * improvements to the editing of minigames * can add multiple lines at once now * can open quick link to google translate with all stimuli text. ## v0.1.1 * implemented manual text minigame creation/editing * NOTE: I'm starting to change the minor version number when I intend to publish to the internet, instead of just publishing pkg to npm. ## v0.0.98 * implemented timer * like most things, is a sketch * really highlights the transience of a minigame. we have to walk the entire history of the minigame ibgib timeline to get "instances" of minigames. This is a poor design, but it's what I got right now. * at the very least, the little numbers increment while playing and not while paused, although the stats shown at the end of the game have an elapsedMs that does not account for pauses. * implemented edit stimuli on minigame * "deprecated" the existing addStimuli api function. * had to implement prune stage for agents when loading when there is a deprecated function * there is some race condition in the agent's timeline that existed in testing for this. the only difference in the agent is timestampMs. * I went ahead and just console.warned when there is a divergent timeline for the agent instead of throwing an error. probably have downstream effects, but there ya go, I'm all alone here. Nobody else using it and no one else damn sure developing. * did basic testing on add, insert, edit and delete stimuli, i.e, not extensive. it looks like it's working and that the agent knows how to use it. ## v0.0.95 * tweaked new project to enter name of new project * meant to obviate issues with renaming the project with agents. ## v0.0.93 * converted left panel projects to projects-explorer * filter projects ## v0.0.91 * Config button added to the shell's header * initial button for clearing/setting (and reading a mask of) the Gemini API key implemented. Now you can see the masked API key, clear it, and update it. ## v0.0.88 * CorpusAnalyzer in the works * collab with Gemini 2.5 Pro * started with analyzeText to build a TfIdfCorpus class * does TF-IDF stuff on text, normal stuff * Moved on to AnalysisEngine * abstracted the concept of a "word" to a "construct" (λόγος), to enable "words" that represent regex-able concepts such as separable verbs for german, dative prepositions, etc. * so now we can TF-IDF and other analyses on texts using text-based `ConstructRule` * name, description, pattern props - all strings * this means... * we can encapsulate higher-order concepts (`ConstructRule.pattern` is composable) using ibgibs themselves. * agents can evolve rules at runtime as needed, persist them, share them, etc., without having to recompile. * TF-IDF and other analyses can be built against these constructs * limited testing done on German, Italian, Spanish, and Koine Greek, but all "tests" passing. * need to eventually turn these to proper unit/integration tests * `CorpusAnalyzer` is the resulting abstraction that wraps the `AnalysisEngine` * Need to write code that consumes `CorpusAnalyzer` in... * typing game component * APIFunctionInfo for use by agents ## v0.0.86 * Minigames are "working", albeit on the bare side. * Project agents can now change the theming (updateCSSVariables function) * Added a blog entry on the new ai-theming and project functionality * Raw view update to enable accordions with summary, separate ib/gib/data/rel8ns sections * Rel8ns now enable navigation, i.e., opening the ibgib as an open tab in the project's center panel. ## v0.0.69 * WIP :under_construction: working on typing/minigame architecture * leaving off having stubbed minigame and typing components, some of the ibgibs themselves, but nothing remotely finished. * juuuust sketched the typing component. the idea is that the minigame component will inject the typing component. other variants of the typing component may be contained in descending classes or just within that component. i dunno, need a break. * WIP :under_construction: already did some plumbing on import, but ran into the realization that this is essentially a timeline merge/sync and thus is non-trivial. is probably a good place to incorporate sync from AWS sync space to just be able to work with a local space. ## v0.0.64 * Export (NO import yet) implemented. * export project, active ibgib tab, each with or without compression. * import will be next * Stopgap hack to restore the most recent theme (i.e. css variables) on app init. * some UI tweaks. ## v0.0.60 * "Files" and persistence "working"! * Project settings are implemented to where the activeTjpAddr (active tab) is saved, as well as open tabs (also by their tjpAddrs). * getIbgibs, mut8IbGib, editProject, editText commands are implemented for project agent. * getIbGibs enables the project agent to get multiple ibgibs, e.g. all ibgibs in the project's "past" rel8nName. * also has a getLatest flag * mut8IbGib enables agents to do fancier (but riskier) mutations on an ibgib/timeline. * editProject gives a simpler interface to editing a project * does things like mutating the project's ib field when the data.name is edited, as well as takes care of timestamps. * editText similarly gives a simpler facade to editing comment/text ibgibs. * used by the text-editor * text-editor component has a debounced saving mechanism now. ## v0.0.55/56 * saving text-editor text via debounce (ty Gemini) * also small busy indicator when this is happening. * need to fill out the UIinfo settings. ## v0.0.54 * just confirmed the refactored code is working. now we have generalized code that can couple local ibgibs with domain ibgibs. this is what we will use to couple local settings to the project ibgibs (and project child ibgibs). * specifically, project ibgibs will need some persistence as to what tabs are open, and what components are representing the ibgibs in those tabs. This is related to my UXInfo or UIInfo data, and the beginnings of our SettingsIbGib_V1. and this is what I am leaving off at. ## v0.0.53 * working on generalizing agent indexing to a "coupled" index. * the use case is to also have "settings" objects that will remember settings. * the driving "settings" config is for remembering what UI component is open for a given project child ibgib. * project component much improved * lens bar * raw and text-editor components are sketched somewhat ## v0.0.43/44 * working on correcting bugs caused by brittle special ibgib index updating. * the (proximal) problem is that we're updating the agent index special ibgib too quickly in multiple calls. So we're getting divergent timelines. * so i'm reworking to use the command service in order to ensure that we have updates to special ibgib indexes happening serially and not in parallel. basically this is using the commanding service act as a poor man's critical section. it still won't defend against multiple tabs, but that is for another (more highly funded) time. * one of the ways I'm doing this is by enabling anonymous function commands. * I'm adding `anonApiFn?: APIFunctionInfo;` to `EnqueuedCommand` and a command helper `commandifyAnonFn` * timeline-api starting dev... ## v0.0.30-40 * Context ibgib workflow is working... * for the web 1.0 agent * a bit for project agents * By this, I mean that no longer do we do the naive chat with an agent, rather, agents subscribe to contexts and react to new ibgibs published to the metaspace pubsub. This is a huge improvement architecturally. HUGE. * Need to start working on project ibgibs actually doing something. * Need to persist more projects. possibly start using tags. ## v0.0.30 * in progress: hooking up agent with context ibgib instead of chat log kluge. * I had to refactor a bit to avoid circular references. * Now agents can be gotten from the agents service by the agent id. * tell user command and hello world command are up and going somewhat. * the hello world mutates the agent's data for the name. * when web 1 component is initialized, it sets the active context of the primary agent. * project agent yet to be done. * tell user implementation not done. * agent is not hooked up to context changes yet (wrt pubsub) ## v0.0.27 * chronology now showing the correct child * very slow initialization, hard to tell right now. * commented on idx forum, may have to transition. * https://community.firebasestudio.dev/t/extremely-sluggish/10502/5?u=ibgib * need to get tab color working * need to associate user with addlmetadatatext on the comment, also timestampinticks * chronologys should initialize children with existing children in addition to handling new children coming down the pipe. * probably should have ibgibs backing it or possibly entire components, hard to tell. (currently i am just throwing up a chat entry like the original chat log kluge) ## v0.0.23-26 * working on chronology (chat log) * in the middle of getting web1 functionality back to "working". of course, the chat log as a previous kluge that we are now fixing. the web1 component creates its own ibgib context now per session (via global storage). after init that as a live proxy, it inits the input to use it and the chronology on the right as well. the input now relates a comment ibgib to that context. * I'm leaving off where I need to hook up a new child to the context in the web component to propagating to the agent. Then I need to continue with the chronology showing the chat, including these newly added comments. ## v0.0.22 * corrected some agent initialization and input bugs. seems to be working better. ## v0.0.21 * created an "agentsSvc" singleton that treats each agent (witness) as a singleton service. It _seems_ to be working, but we'll see. At the very least I'm able to just chat with the website again, as well as change the theme. ## v0.0.20 * changed to input component * working on breaking out more agents. * still working on this. the primary agent is now kind of a website agent. But for some reason the agent doesn't have those system instructions that describe the ibgib protocol, website, etc. * UI init changes for drawing the eye better and more formalized * leaving off before changing the asides. I should script those asides with highlights I think. ## v0.0.13 * added dynamic theming backed by agent * still need to fill out more css variables * ibgib projects * instead of the raw ai canvas * work in progress (incomplete) * created project api command related things but created a dependency order error. ## v0.0.12 * first funding! added some info to the funding page. * Added blog and faq * tweaking layouts/css * path renderable improved * added quite a bit of instructions for interacting with the primary agent. * anticipating users chatting with the agent about the protocol and the website. * other things as well. * will have an ai look at diffs when ibgib versioning going, which will be * easier with ibgib vs git IMO, but we'll see. ## v0.0.8/9 * path renderable implementation started. ## v0.0.7 * simple send button for input textareas for agents ## v0.0.6 * fixed the pesky "race condition" of indexed db storage initialization. * was due to a logical error on my part when calling an inner `initializeStorage`. I had to explicitly call db.close, as a final db.close was not sufficient for the outer function, `storageCreateStoreIfNotExist`. * fixed db.close errors throughout the storage helper functions. * refactored storage functions into a separate file outside of the more general `helpers.web.mts`. ## v0.0.5 * I'm making the default local user space have an automatically-created space name to avoid prompting the user for the space name. * This is due to the prompt dialog screwing up the first user first land on ibgib.com experience. I want the user's eye to be drawn to the aside on the home.html component. * I've added some plumbing to make the bootstrapping of the blank gib app idempotent. * This was in thinking that we may try to bootstrap later, but with the auto-local space naming, i'm holding off on this.