All Issues
24,993 verified issues
[BUG] Failed to install npm package from git in docker since v6.11.0
What / Why This bug is related to https://github.com/npm/cli/issues/476#issuecomment-553917541 and https://github.com/npm/cli/issues/514 When `npm install` packages from git in docker Where packages from public git repositories How Current Behavior [code block] Steps to Reproduce Create simple package.json with git dependency [code block] run `docker run --network=host -it -v $(pwd):/var/work/test -w /var/work/test node:12.14.0 /bin/bash` `ls -l` shows that files are owned by unknown user and unknown group (this is typical behaviour on build agents like teamcity, teamcity clones git repo and mounts it inside docker container) [code block] run `npm install` [code block] run `npm i -g npm@6.10` (downgrade to npm 6.10.X) run `npm install` [code block] Expected Behavior `npm install` / `npm ci` works well like npm@6.10
Add support for Cloudflare Workers / Vercel Edge Functions
So far it seems like there are 3 culprits which may block us from running `Langchainjs` on Edge. - [x] 1) OpenAI and it's usage of `axios` #118 - Switch to a different library / create a upstream PR for the official library (the pending MR seems to be doing way too much IMO) - [x] 2) `**/index.ts` files might include other modules, which aren't needed - better treeshaking support, which - [x] 3) `langchain/util/hub` and the general usage of NodeJS APIs (`fs`, etc.) #97 - [x] 4) Have at least 1 vector store which 1. doesn't need Node native modules and 2. stores metadata for each document #73 - [ ] Add tag in the docs for node-only things (eg hnswlib) - [x] Create and test cloudlfare workers and vercel edge templates - [x] Remove usage of node-fetch (which uses node-only APIs) #118 - [x] Remove all usage of require #124 - [x] Get TS to output ESM #124 - [x] Remove usage of node `crypto` #213 - [x] Replace node-only tokeniser with one that works in browser-like envs #214 This is the error output when attempting to deploy the `examples/src/chains/llm_chain.ts` via Vercel: <img width="582" alt="CleanShot 2023-02-19 at 19 28 23@2x" src="https://user-images.githubusercontent.com/1443449/219971669-827be1f6-fd2c-4b3c-995c-b5d572a0847d.png">
[DevTools Bug] Element "9" not found | Also "Element "10" not found" / "Element "12" not found"...
Website or app https://github.com/arbocobra/redux-minesweeper Repro steps On first load I get multiple Error warnings on all app components from `<Game/>` down. They are identified with different Element numbers but appear to be directed to the same issue "The error was thrown at chrome-extension://fmkadmapgofadopljbjfkapdkoienihi/build/main.js:39558:15" If you select 'Begin Game' in app the error in component` <Game />` can be dismissed but new Errors appear on all newly rendered components below (once again all have different Element numbers directed to the same location (../main.js:39558:15) // This is issue is probably related to [[DevTools Bug] Element "41" not found](https://github.com/facebook/react/issues/25755) Chrome is up to date (Version 108.0.5359.71 (Official Build) (x86_64)), and restarted. I also tried reinstalling React extension (running 4.27.0). The app is using Redux, which one commenter in other thread mentioned might be a commonality. How often does this bug happen? Every time DevTools package (automated) react-devtools-extensions DevTools version (automated) 4.27.0-bd2ad89a4 Error message (automated) Element "9" not found Error call stack (automated) [code block] Error component stack (automated) [code block] GitHub query string (automated) [code block]
Is this package still actively maintained?
Many people and organizations rely on this package, but the last commit was made in September 2023. The list of open PRs is also growing, with most of them left with no review from a collaborator, let alone merged. Now that @luin is no longer the primary maintainer of `ioredis`, who can we rely on for ongoing maintenance? Who is reviewing issues and PRs? Do we need to expand the list of collaborators?
Provide precompiled binaries for ARM
Currently, we do not have any precompiled binaries for the ARM platform. Now that Windows is supported, we should look into compiling for ARM. That would enable users to Prisma 2 on Raspberry Pi and other all sorts of ARM-based devices. This will also include adding ARM to the binaries spec: https://github.com/prisma/specs/tree/master/binaries#binary-files Ref: https://github.com/prisma/prisma2/issues/132
Axios returns ECONNREFUSED 127.0.0.1:80 but port is set to 8080
Describe the bug Axios returns ECONNREFUSED 127.0.0.1:80 but port is set to 8080 To Reproduce While having an express server returning 200 on http://localhost:8080/ (and confirmed via browser), running the following returns the error provided (see additional context) [code block] Expected behavior Axios sends an http request to the port specified in the url Environment Im indirectly using axios@0.21.1 (start-server-and-test -> wait-on -> axios@0.21.1). - Axios Version [0.21.1 - Adapter HTTP - Node.js Version v12.8.0 and v15.8.0 - OS: macOS 10.15.6 Additional context/Screenshots Error: [code block]
feat: v11 goals & suggestion box
> 🚀 v11 is available to use and ready for production: https://trpc.io/docs/migrate-from-v10-to-v11 Summary The main goal of v11 is to improve our own quality-of-life by removing v9-support as it makes the codebase very messy. Work will be done in the `next`\-branch Ideas welcome from the community of API-decisions you think we should revisit. We plan not to delete anything which hasn't previously been marked as `@deprecated` (if we rename, we'll do an alias) Discord chat: https://discord.com/channels/867764511159091230/1057652120473575505 Plan Before official alpha [X] Delete v9 support #3494 [X] Recover test coverage from the above -- it dropped down to 93% -- see https://app.codecov.io/gh/trpc/trpc/pull/3494 [x] Deprecate and rename: Remove `Proxy` from naming of functions. We have `Untyped` that is for the untyped. e.g: `createTRPCProxyClient` -> `createTRPCClient` #4932 [x] Look over all `@deprecated` and delete accordingly (#5026) [X] Move from using string paths to tuples in `react-query` [ ] Look if there are any open issues that we want to do that might entail breaking features [ ] Consider if we wanna do them or wait for v12 - maybe we start exporting some stuff specifically for plugins [x] Turn `TRPCError.cause` to `unknown` [x] #5188 [x] Once we've created a `@trpc/core`-package, we need to look at extensions using tRPC and make a way to extend tRPC for libraries - #5308 [x] #5210 [ ] Make sure things that are exported in `@trpc/client
[expo-updates] [SDK 49] Presence of expo-updates causes production build to crash on launch
Minimal reproducible example https://github.com/mikescott/expo-updates-crash Summary The presence of expo-updates causes production and simulator builds to crash on launch. I discovered this after upgrading a project to SDK 49. The issue does not happen in SDK 48. To reproduce 1. Build the app: `eas build --platform=ios --profile=simulator` 2. Run the app in an iOS Simulator The app will crash with the following report: [code block] To fix the app: 1. `yarn remove expo-updates` 2. `eas build --platform=ios --profile=simulator` 3. App will launch successfully Environment [code block]
Auth error only after recaptcha: "Cannot read property 'style' of null"
Describe your environment Operating System version: Linux 5.7.5-arch1-1 x86_64 Browser version: Firefox 79.0b3 (64-bit) Firebase SDK version: 7.15.5 Firebase Product: auth Describe the problem I am setting up Phone Auth with firebase. If there is a problem with the phone number (too long, too short, etc), but the captcha is shown (delaying the error until after resolving the captcha), the following error is thrown: [code block] I have a feeling it happens when firebase auth attemps to reset the recaptcha after a problem with phone auth occurs. There is no issue when captcha is shown but phone number is valid. Steps to reproduce: Go to the below codesandbox (open it in incognito -- hopefully that'll make sure captcha is shown) Open console Click submit (with an invalid phone number) Ensure captcha is shown, if not, refresh and try again until it is Observe the following error is thrown: [code block] Relevant Code: https://codesandbox.io/s/compassionate-bouman-vw5zn?file=/index.html Related issues: https://github.com/dozoisch/react-google-recaptcha/issues/129 * https://github.com/google/google-api-javascript-client/issues/281
PostgreSQL autovacuum not triggering on large high-churn table causing query slowdown
Queries on a heavily updated PostgreSQL table degrade over time despite indexes being present. EXPLAIN ANALYZE shows sequential scans. pg_stat_user_tables shows n_dead_tup in the millions and last_autovacuum is hours or days ago. The default autovacuum_vacuum_scale_factor of 0.2 (20%) means autovacuum waits until 20% of the table is dead tuples — on a 50M row table that is 10M dead tuples before autovacuum triggers.
Expo EAS build fails: runtimeVersion policy not supported in bare workflow
Expo EAS build fails for projects in the bare workflow when app.json uses `"runtimeVersion": { "policy": "appVersion" }`. The policy-based runtime version is only supported in managed workflow projects. Bare workflow projects must set runtimeVersion to a literal string matching the version field. AI-generated app.json files and Expo documentation examples often show the policy form without this distinction.
JWT refresh token silently corrupted when renewal endpoint omits field
After token renewal, the app stops working silently. The renewal API returns only { token } but the client code does `const { token, refreshToken } = response.data` and stores both — meaning refreshToken becomes undefined and overwrites the valid stored refresh token. The user appears logged in but all subsequent authenticated requests fail until they manually log out and back in.
Prisma "prepared statement already exists" error in serverless environments
Prisma throws 'prepared statement "s0" already exists' in serverless functions (Vercel, AWS Lambda, Netlify). Prisma uses named prepared statements for performance. In serverless, multiple function instances may share a PostgreSQL connection via a pooler. When a new invocation reuses a connection that already has a prepared statement with the same name from a previous invocation, PostgreSQL rejects it.
Expo React native new EAS build blank white screen
Summary When i Created a APK build using EAS it just shows me white screen and there is also no logo of my app after installing meanwhile when i use in locally using Expo app the app is working perfectly fine and my app also have splash screen and logo. This is my eas.json { "cli": { "version": ">= 7.1.2" }, "build": { "development": { "developmentClient": true, "distribution": "internal" }, "preview": { "distribution": "internal" }, "production": { "android": { "buildType": "apk" } } }, "submit": { "production": {} } } What platform(s) does this occur on? Android SDK Version "expo": "^51.0.10", Environment expo-env-info 1.2.0 environment info: System: OS: Windows 10 10.0.19045 Binaries: Node: 21.6.2 - C:\Program Files\nodejs\node.EXE npm: 10.8.1 - C:\Program Files\nodejs\npm.CMD IDEs: Android Studio: AI-231.9392.1.2311.11330709 npmPackages: expo: ^51.0.10 => 51.0.10 react: ^18.2.0 => 18.2.0 react-native: 0.74.1 => 0.74.1 Expo Workflow: managed Minimal reproducible example NA
Feature proposal: `children` variant
The idea would be to be able to do things like this: Input CSS [code block] Output CSS [code block] Note: All `children:` variants would need to be output before the normal rule-sets so children can override rules defined by their parent (`.class` and `.class > *` have the same specificity). HTML [code block] Advantages - Less repetition in the HTML (easier to change things that should be changed together) - Easier to read - Smaller HTML file size I think this would be particularly useful for grids and lists. What do you think?
should I use express for future projects?
express was the most promising web server for node.js, it seems the development has been slowing down considerably for the last few years, as of today ES6 is not added, no async/await work in sight either, this is quite a surprise considering the (still) fast advancements on frontend frameworks, node.js and javascript. What is going on? Maybe it's time for a fundraise to re-ignite the development for this great project? or is there something else replacing it so nobody cares that much? or node.js framework is losing its popularity in general after some over-hyper? Really want to see a django alike framework in node.js.
Unable to create Partitioned cookie
The Partitioned attribute is not set when passing the option to res.cookie(). https://developer.chrome.com/docs/privacy-sandbox/third-party-cookie-phase-out/#partitioned-cookies
Badge always incorrectly displayed as "failing"
This badge is shown as "failing" but the tasks' statuses are all green. Describe the bug The badge was green since CD pipeline is prechecked before running, meaning it normally should never fail. The badge for it was green and it suddenly went to red - "failing", even though all constituent tasks are succeeding. To Reproduce See 1st paragraph. Expected behavior The badge is green when all tasks successful, red when at least one fails. Runner Version and Platform Full yaml snapshot available here. Simplified yaml: [code block] OS of the machine running the runner? Linux
`@apply`Broken in Tailwind CSS v4.0 – No Clear Fix or Docs!
Tailwind CSS v4.0 `@apply` Bug Report Introduction I'm in the process of upgrading a large project, actually, five Blazor projects that all conjoin into one via components and different backends. Our focus with this project is to deliver an optimal UI experience to the end-user, which is why we use TypeScript, C#, and of course, our favorite, Tailwind CSS. Btw, huge fans! Now, with that said, we have certain native components where we cannot add CSS classes to the component itself, making `@apply` an essential part of our workflow. However, after upgrading to Tailwind CSS v4.0, `@apply` is no longer working as expected, and there seems to be little guidance on how to transition projects that rely on it. This bug report aims to outline the issue clearly and request either a fix or detailed documentation on what exactly we need to do to overcome this. What version of Tailwind CSS are you using? v4.0.0 What build tool (or framework if it abstracts the build tool) are you using? Webpack 5.97.1 What version of Node.js are you using? v20.11.1 What version of npm are you using? v10.7.0 What IDE are you using? Rider IDE What operating system are you using? macOS Reproduction URL There's an active discussion on this issue where many developers are experiencing the same problem but no one has officially reported it yet. Rather than debating, I’d rather get a solution, so here’s the link to the discussion: https://github.com/tailwindlabs/tailwindcss/discussions/13336#discussio
Prisma query engine binary crashes on Amazon Linux: binaryTargets mismatch
Prisma works correctly in local development (macOS or Debian-based CI) but crashes on production Amazon Linux (Elastic Beanstalk or Lambda). The error is "Query engine binary for current platform 'rhel-openssl-3.0.x' could not be found". Prisma ships platform-specific binary engines and the schema.prisma must list all target platforms in binaryTargets.