如何让Heroku识别一个子目录（不是root）中的yarn.lock或package.json

问题描述

我有一个Rails应用程序，使用React，Webpack和Yarn作为客户端。我在/ client目录中拥有与客户端相关的所有内容。这包括我的yarn.lock和package.json文件。我有一个Procfile，它可以在/ client中执行cds并正确运行 yarn run build 。这在本地工作，没有错误。但是，当我部署/推送到Heroku时，我的推送被拒绝，并且出现以下错误：

 远程：cd client& amp ;&安培;纱线运行构建：生产
远程：sh：1：纱线：未找到
远程：耙中止！ 
 remote：命令失败，状态为（127）：[cd client&&生产线：生产...] 
  

这对我说的是Heroku不能在根级找到yarn.lock文件。如何让Heroku在my / client目录下使用 yarn.lock 文件？

或者是原因这个错误的其他东西完全是什么？

解决方案

刚刚通过类似的事情我自己。您需要订购构建包，以便节点处于第一位。如果一个yarn.lock在你的根目录下（或者你的引擎键在@ remydib的答案中），那么节点buildpack将安装纱线，然后安装你的包。如果你在/ client中有一个 yarn.lock ，把它放在根目录下，然后在它上面添加一个 postinstall 脚本，说 cd client&&在我的情况下（rake资产：预编译调用纱线并失败），出于某种原因，当buildpacks进入ruby时 - >

节点，Heroku在尝试运行rake任务之前不会等待节点安装，我们会去kablooie。

希望这有助于您。

I have a Rails application using React, Webpack, and Yarn for the client side. I have everything relating to the client side within the /client directory. This includes my yarn.lock and package.json files. I have a Procfile that cds into /client and properly runs yarn run build. This works locally with no errors. However, when I deploy/push to Heroku, my push is rejected and I get the following error:

remote:        cd client && yarn run build:production
remote:        sh: 1: yarn: not found
remote:        rake aborted!
remote:        Command failed with status (127): [cd client && yarn run build:production...]

What this says to me is that Heroku wasn't able to find a yarn.lock file in the root level. How do I get Heroku to use the yarn.lock file within my /client directory?

Or is the cause of this error something else entirely?

解决方案

Just been through a similar thing myself. You need to order the buildpacks so that node is first. If a yarn.lock is in your root dir (or yarn is in your "engines" key as in @remydib's answer), then the node buildpack will install yarn and then your packages. If you have a yarn.lock in /client, put one in root, and add a postinstall script to it that says cd client && yarn run build.

In my case (rake assets:precompile calls yarn and fails), for some reason when the buildpacks go ruby -> node, Heroku doesn't wait for node to install before trying to run rake tasks, and we go kablooie.

Hope this helps.

这篇关于如何让Heroku识别一个子目录（不是root）中的yarn.lock或package.json的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！