use consistent chunk ids #1534
use consistent chunk ids #1534
Conversation
truncate the md5 of the chunk name to a 64 bit integer
|
Deploy preview failed. Built with commit a63f1a7 https://app.netlify.com/sites/using-styled-components/deploys/596cfd386f4c5075cc3c3b5f |
|
Deploy preview ready! Built with commit a63f1a7 |
|
Deploy preview ready! Built with commit a63f1a7 |
|
Deploy preview ready! Built with commit a63f1a7 |
|
Hey @stevensurgnier doesn't the |
|
The webpack-md5-hash produces the chunkhash - a hash of the chunk's content. This PR modifies the chunk id - an internal webpack id. By default the chunk id is represented as a sequential integer. You can see it in the manifest and a final chunk file. The problem I ran into is that the chunk id can change when you add more chunks. However, the chunkhash is a function of the chunk content not the file content. Thus, the file can change while the chunkhash remains the same. If you're caching based on chunkhash (or filename) this will cause big problems. This PR provides a consistent chunk id based on chunk name. Unfortunately, the version of webpack we're on requires the chunk id to be an integer. Thus, the not so ideal amount of entropy used in this PR. |
|
This is awesome! Thanks for researching the fix and creating the PR. |
|
@stevensurgnier, Hmm I didn't consider that, super interesting thanks for the explanation 👍 |
Consistent chunk ids are generated by truncating the md5 of the chunk name to a 64 bit integer.