On Thu, Dec 02 2021, bill-auger wrote: > On Wed, 13 Oct 2021 14:32:54 +1100 Yuchen wrote: >> based on the ethical >> repository criteria[1] it would probably get an F for failing C0, >> because with LibreJS on I couldn't register a new account, nor >> could I create a pull request. > > pagure deserves some defense - the pagure devs have stated > willingness to accept any and all patches for librejs - > pagure.io logins are managed by fedora though - those are the > folks who would need to be convinced to adapt for librejs, if > pagure.io fails C0 for that reason > > the C0 question is essentially: > * does any important functionality require non-free software? > strictly speaking, a complete API satisfies C0, naturally; > but not for the registration process > > example of important functionality: > > * can i register with librejs? No.
In fact, I can. I was able to register it in eww (the emacs built-in browser without any javascript capacity), but then I couldn't log out. One can probably register with noscript on. This is why I think LibreJS should have a noscript mode, like the opposite of whitelisting, where the user can block all javascript (trivial or not) on a website. Without more research I'm not sure if this is doable though, as I observe in the case of noscript there's a "disable noscript for this tab" button, but not "enable it for this tab". It is a pain to have both librejs and noscript on, so whenever I want to test whether a site works better noscript than with librejs, I would test it in EWW. Discourse forums are another example. > > IMHO, C0 is satisfied if either of the following such questions > are true of each important feature (register, open a PR, etc) > > * can i ___ without JS, or without a web browser? Yes. > the pagure API supports authenticated pull-requests > it is a (non-web) interface to all important features > (using the 'curl' web browser) > https://pagure.io/api/0/#pull_requests-tab Thanks, I will test the API. > > * can i ___ with librejs in a web browser? No. > normally, the cause can be upgraded remote dependencies, > but the librejs licenses table(s) were neglected - a routine > maintenance task - the fixes would be tiny and accepted > > in short, i dont believe that C0 requires all important > features to be exposed via the web interface - a complete API > makes C0 a moot point, naturally Perhaps this has been discussed in the repo-criteria-discuss, but I can see why this argument makes sense, in that a user seeking freedom has the means not to give it up while still being able to use the service. > > a trivial example: the VCS server - the VCS server is an > important feature (if not THE most important) - the forge is not > expected to expose all functionality of the VCS server via the > web inteface, nor to mimic a VCS client - i interpret C0 > broadly, enough to allow any of all important functionalities > to be satisfied by any mechanism operated by the same host - > it is not important to me if all functionality is supported via > the web interface, as long as it is possible, using some libre > tool (eg: the 'git' or 'curl' programs) > Best, Yuchen -- PGP Key: 47F9 D050 1E11 8879 9040 4941 2126 7E93 EF86 DFD0 <https://ypei.me/assets/ypei-pubkey.txt>
signature.asc
Description: PGP signature