What is the correct way for a web service to provide FET’s source under AGPL?

[ogursoy]

Hello,
I have a general licensing question that I could not find a clear answer to in the forum archives.

There are a number of web services now that run the FET command-line engine on a server behind their own web interface. The user uploads or enters data, the service runs FET, and the resulting timetable is shown in a browser.
For that arrangement, what does the FET project consider correct practice under AGPLv3?
   

• Is a link to the official download page enough for an unmodified binary, or should the exact source archive matching the running version be hosted by the service itself?
• Should the FET version number be shown to users? "Corresponding Source" does not seem very meaningful if nobody knows which version is running.
• If only the command-line entry points are patched — messageboxes.cpp or fet.cpp, which the README itself suggests for programmatic integration — does that count as modifying the Program for the purposes of section 13?

I ask because I have seen several of these services link to a GitHub repository for FET source, and as far as I can tell no official FET repository exists on GitHub. The authoritative source appears to be the tar.xz archive on lalescu.ro, so those links seem to be written from memory rather than checked.
If there is a preferred wording or a standard set of links you would like such services to use, it would be very useful to have it stated somewhere public.

Thank you for FET.

[Liviu Lalescu]

Dear ogursoy,

Thank you for your interest!

I do not work incrementally, with GIT or SVN or CVS, I never did and I have no such knowledge. But I maintain a GitHub page: https://github.com/lalesculiviu/fet (only the official tar.xz archives).

It was @Volker Dirr  who advised me that we choose GNU AGPLv3. He knows better than me. So, I invite him to answer your questions.

You are welcome! Thank you for your kind words!

[Volker Dirr]

Well, I am not a lawyer. So far I know:

1) In my opinion a link is already fine. The "only" problem might be that in worst case you want to (or need to) to check/prove if you really didn't modify the FET source. It will be easier if guys know the exact version so they can check easier. So far we luckily stored nearly all released versions, but they are not all public available anymore. Many are still available on source forge. Nearly all version from the last ~20 Years are stored on my webserver (public available, but i remove the link some time ago, since I saw some guys always downloaded old versions. But in fact, if you know the link, you can still see all files). And as far as I know Liviu has also all versions stored private (like me). So for "safety" I think it is better to add such an download to the used version. That is only ~15 MB, so no big problem.

2) See 1). We will be glad if we are "supported" by a link (even without the version number). If you want to check / need to prove that the source is unmodified, then is much easier if you know the version. So a version will help much. In my opinion it is even clever for those guys to print the version number, since users can see that the software is up to date and not an old one. (even i think there is no "law" to print the version. But I am not 100% sure.)

3) Well in that case the AGPL is clear: If you modify the source, so even if you just add a single " " (space) in a comment, that doesn't effect the binary file, even then you modified the source and then you need to publish it.

[ogursoy]

Dear Liviu and Volker,
Thank you very much for your time and beautiful software FET.
Best

[Liviu Lalescu]

You are welcome, ogursoy!