Closed IlCallo closed 3 years ago
Regarding:
pug-plain-loader (used in docs) not updated since years, webpack 5 status support unknown
I'm using a forked version because as you noted, it hasn't been updated for a while. The proper version by Evan has a dependency error and doesn't work with pug v3. See Other Packages in the forked readme for details.
This only depends on the latest version of loader-utils
by webpack, so I believe it will be compatible.
I can test/confirm once ready to review.
@Blfrg thanks, updated into the initial post :)
@IlCallo just a little update: workbox-webpack-plugin v6(stable) is released, and the latest version is v6.0.2.
Perfect, updated into the first post
@IlCallo There is a temporary fork of postcss-rtl that's been upgraded to work with latest postcss plugin structure. https://www.npmjs.com/package/@mjhenkes/postcss-rtl I am thinking we may need to fork that (the form) ourselves and maintain a @quasar/postcss-rtl
I added this to the original post, but I'm not sure taking the burden of yet another package to maintain for ourselves is the right way...
I agree. I am using rtlcss
, but Razvan says that doesn't handle all of our cases properly. It may be better to PR that package with what's needed. Right now I can't say I know what is missing from rtlcss, but it is actively maintained.
FYI the vue fork of preload-webpack-plugin
has supported webpack 5 https://github.com/vuejs/preload-webpack-plugin/releases/tag/v2.0.0
I've also maintained a fork of friendly-errors-webpack-plugin that has supported webpack 5 https://github.com/sodatea/friendly-errors-webpack-plugin/releases/tag/v1.8.0 (@soda/friendly-errors-webpack-plugin
on npm)
Thanks! One or two out of the list then :D
Regarding:
pug-plain-loader (used in docs) not updated since years, webpack 5 status support unknown. Consider using this updated fork instead
I was just reviewing my dependencies and noticed that back in December @sodatea merged the needed PR
for the orginal pug-plain-loader
to work with pug v3.
So the original dependency can be considered reliable again, instead of the fork.
Thanks, updated in the initial list
html-webpack-plugin v5 has been upgraded to the latest stable version long ago .
Just as a note, as vite
support is also on roadmap: vite brings significant performance compared to webpack. After trying it, you might consider making vite a higher priority than webpack for 2.0 (speaking only from my personal experience; there could be roadblocks to using vite for other projects). Thanks for considering.
We need to first concern about our current user base and not leave them hanging with their current apps. Due to Webpack team not upgrading chokidar
dependency for Webpack 4 we can't have apps correctly running on Node 14+ (which is LTS!).
We are aware of how good Vite is, which is why we've planned a Vite plugin and even a Quasar CLI based on Vite instead of Webpack. It will come, guaranteed. I'm about to finalize the Webpack 5 upgrade (a few more days and there's going to be a q/app release with it) and then it's Vite time!
Mulțumesc, Razvan :)
"@quasar/app" v3.0.0-beta.16 will feature Webpack 5. Was hell upgrading everything (including our many of custom webpack plugins). But while I was at it, took time to also greatly improve and revamp the inner workings of q/app (and there are so many to list them here!). Get ready to rock'n'roll your apps with it!
Available through "@quasar/app" v3.0.0-beta.16 + Quasar v2.0.0-beta.13. Enjoy and please read the release notes! https://github.com/quasarframework/quasar/releases/tag/%40quasar%2Fapp-v3.0.0-beta.16
https://webpack.js.org/blog/2020-10-10-webpack-5-release/ https://webpack.js.org/migrate/5/
webpack 5 is out and it's time we go way down into that rabbit hole :D
We'll use this issue to keep track of everything. We plan to ship Quasar 2 with Webpack 5 support, or add it right after with a
@quasar/app
update, ~but could push back the upgrade to Quasar 3 if we find out there are too many breaking changes for devs~ this is needed as some webpack 4 issues are pretty severeIf anyone from team or community wants to jump in and point out something we are missing, both on current issues or new features we can leverage, please do :)
Let's avoid spam and "+1", we are going to delete/hide those comments
Existing issues due to webpack 4
New feature we want to leverage from webpack 5
??
andexport * as ...
. This should allow us to avoid using babel in some cases@types/webpack-*
packages and reduce conflicting types problemsimport {date} from 'quasar; const { ... } = date
syntaxLoaders to upgrade
writeToDisk
option. This dependency should be removedIssues with webpack 5 migration
postcss-safe-parser
due to missing API oncss-minimizer-webpack-plugin
Related issues
Things which could be explored, but which aren't really "in-scope" for the migration
preload
support is still blocked, as both the upstreampreload-webpack-plugin
and the forked@vue/preload-webpack-plugin
are pretty much dead. We could ask the Vue team if they plan to update their fork, as I don't think the original repo will ever come backAdditional context Vue CLI still has not switched, but something is moving:
React is moving there too and have an exaustive list of stuff to upgrade and problems to solve:
Angular CLI already supports Webpack 5 as experimental (we could check if we can learn something from their code), and is discussing about Module Federation feature: