Arch Linux

> you should not `ln -s` a file from another package in `browserpass-firefox` `browserpass-chromium`, instead, you should install to the right place for firefox or chromium in `browserpass`, and the `*-{browser}` should only contain the extensions rather than native-host included

Why? The official upstream Makefile installs com.github.browserpass.native.json to /usr/lib/browserpass/hosts/{firefox,chromium}/, why manually deviate from that?

> the Makefile isn't required for browserpass to work. please remove it

Again, that's an issue with the upstream Makefile installing itself to /usr/lib/browserpass/ for whichever reason, not a packaging issue, the PKGBUILD only executes a standard "make install".

If you want to get these peculiarities fixed, I suggest opening an issue upstream at https://github.com/browserpass/browserpass-native to avoid every distro packager having to work around them manually, the Arch PKGBUILDs for browserpass look as expected.

> you should not `ln -s` a file from another package in `browserpass-firefox` `browserpass-chromium`, instead, you should install to the right place for firefox or chromium in `browserpass`

I disagree, `browserpass` is a native extension, it should not assume which browsers user has installed and should not assume that user wishes to use the extension in all browsers. However because of how native extension protocol is implemented in browsers, the json files _belong_ to the native host app, for example because they depend on the location of native host executable on disk. Therefore, they are installed in browserpass directory, and user is expected to symlink to them using `browserpass-{browser}` extension or manually.

> Makefile isn't required for browserpass to work. please remove it.

It is required to properly configure the extension. The `browserpass-{chromium,firefox}` performs the magic for Chromium and Firefox browsers, but for all other browsers users are expected to use this Makefile to configure the browser.

If you wish to discuss this further, please do open a ticket on Github and I would be happy to follow-up.

Installing the native messaging protocol whitelist does not "assume that user wishes to use the extension in all browsers". It merely permits it *iff* the user does in fact install the extension in said browser, while also making it work even if the user installed the extension to their $HOME profile rather than with pacman.

True but I do believe it is better to let this package install json files in its own directory and let users choose where to symlink json files (with the help of packages like `browserpass-{chromium,firefox}`, provided Makefile or manually), rather than pollute file system by installing json files for every possible browser, even the ones that user doesn't have installed, and even the ones available only in AUR such as Google Chrome.