Threaded Mode | Linear Mode

3298 · 11-17-2021, 11:47 PM

(11-17-2021 09:22 PM)Claudio L. Wrote: I think there's no doubt the license of x49gp cannot be anything other than GPL because of being a derivative work of QEMU.
All contributors knew this when they worked on the project, so I don't see why you would need to contact them one by one. Everybody knows that contributing to a GPL project means their contribution will be released under the GPL.

I'm not a lawyer either, but as far as I remember one of the big changes between GPL v2 and v3 was adding many other licenses to a list of those GPL considers "compatible" (which some projects even cited as a reason to stay v2-only instead of v2-or-newer). As far as I can tell, the existence of this list gives us some wiggle room to choose a one of them for those parts that aren't directly QEMU pieces. In that light contacting the contributors who constitute "us" for clarification may or may not be needed, but doing so would definitely be the safe option.

(11-17-2021 09:22 PM)Claudio L. Wrote: Source code is available in multiple places, all you'd need to do is perhaps add a URL to any of the github repos to comply with the GPL. I don't think the license requires you to include the sources in the distribution, but for you to make the source code available, which is a different thing. Pointing to a github repository should suffice (I think, I'm not a lawyer). Otherwise all Linux packages would have to be distributed with full sources. Instead, the package tells you where the sources that were used for the build are located, and you are on your own after that.

Oh, I know that. But I also know what people do with binary packages, and that distressingly often omits the linking to source part. Look no further than x49gp itself, particularly the list of forks on GitHub ... last time I checked (anonymously of course; my stance on their terms and conditions hasn't changed for the better since Microsoft took over), the twenty or so newest commits were from a particularly dumb user who forked your repository, then started removing all the files over many commits (one by one at first, then in batches), then added a binary package and a fresh readme, and changed both of these many times in quick succession.
Aside from the disgusting misuse of git as a version control system, this idiot didn't deign to tell others what sources the binaries were produced from. Sure, the commit history kind of retained the source files, but which revision(s?) were the binaries produced from, and were there custom patches? Nobody knows.

If x49gp already has such an example in its current state, I don't even want to imagine what nonsense people would do if binary packages were pre-made. Given the average human stupidity, it probably won't get better down that path, only worse.

(11-17-2021 09:22 PM)Claudio L. Wrote: As far as the binaries... they should be removed from the repository. Then perhaps create a script that would download the formal HP ROM zip files from hpcalc.org, extract the actual rom files and cleanup.
That leaves only the boot code... that should be moved away to some obscure repository. Maybe not obscure, perhaps a .zip file uploaded to hpcalc.org with the boot code for all supported calcs, then the user can go and download it by hand (or a script provided that does it during install). I believe hpcalc.org has authorization to host binary files copyrighted by HP, so no issue there.

Hey, quarantining HP firmware and bootcode to hpcalc.org and pulling it in during build or installation is a pretty great idea. Eric Rechlin indeed has HP authorization to host the firmware, though I'm not sure if that covers the bootcode too. He currently doesn't offer that for download at least (apart from a copy of x49gp, but that doesn't count).