[mushufasa]

So for complete clarification:

If I hosted a wiki.js, say on a subdomain, unrelated to a commercial application's code at all, but lined to it (e.g. the "docs" section of navigation) would I need to disclose

a) the source code of the wiki.js subdomain only, and any modifications we make to it

b) the source code of wiki.js and the other proprietary app BOTH ?

[dspillett]

A, unless you package and distribute the code of wiki.js in some way with the application - but it sounds like you are talking about a hosted only solution so that won't be an issue.

If you allow self-hosting of your proprietary code by your users and the refers to your documentation server than this would still not trigger AGPL for the other code - you are not distributing wiki.js or parts of it. If you allow self hosting of the other code and include wiki.js for a local copy of your documentation then APGL might trigger.

Linking is meaning software linking, not referencing your own content displayed by the software. Think of it as the same difference between including using chunks of a paper in your own paper, or saying "as discussed in This Paper About This Thing (Him, Her, et. al, 2013)...". Or, as in this case the content is your's, referencing your own paper.

This potential confusion with the word "linking" (as natural languages are fairly dynamic beasts) is why people are sometimes overly fearful of AGPL and GPL (and often even LGPL) software.

[oneplane]

Linking as in binary linking: yes (option b). Linking as in HTML A href link: no, only option a.

Problem is that unless you are installing your servers by hand you end up re-packaging wiki.js and then your own packaging code and the modification on wiki.js as a build artefact are now supposed to be public as well.

[mushufasa]

I don't understand what you mean about packaging code.

Do you mean to distinguish between setting up an EC2 instance or digitalocean droplet type of thing versus using ansible/chef/etc orchestration among the servers? Implying you would have to publish the configs for all your servers?

That would make no sense, both from a practical perspective (publicly disclose server configs? those can include sensitive information), and a theoretical perspective (what does that have to do with the source code at all?).