Hi Mark, I'm a bit in a hurry and do not really want to focus on what happened in Harvey: to my eyes that story just show you cannot trust people just because they are nice and well known "open source" contributors, or because they work for big multinational that "do no evil" or even join the Good Guys (TM) of Software Freedom Conservancy.
But let me clarify On Wed, 31 Mar 2021 13:34:17 +0200 Mark Wielaard wrote: > I looked a bit at that issue you filed and how they handled your > request to remove your code from the project. And I must say I don't > really understand what you believe they did wrong, they seemed to have > acknowledged and corrected their mistake and then removed all the code > you wanted to have removed. I asked them to `git revert` my changes referencing the issue, so that the code I reused in my own fork of Plan 9 was safe that nobody could claim copyright of my work after, say, a change in the version control system adopted by the project. Instead they did a `git rebase` over which, I was pretty surprised actually, they "accidentaly" squashed some of my own commits verbatim (but without my name) in incredibly large commits. And you know, they had to git push -f such rebase, breaking all the existing github forks (while the `git revert` approach would not have caused any issue to anybody) > There is some disagreement over whether a > mass change of function declarations is copyrightable or not. And implementations. And kernel changes that took a couple of days to get right (Harvey kernel was pretty unstable back then). And more I did not remember but I noticed back then: > But I happen to agree with them that if there is only one way to do > it, then having someone else do the same transformation is a correct > way to resolve this. Sure! But first, there were several different ways to do that (several equivalent typedefs were already in place in u.h, without even mentioning macros and so on), and more importantly if you actually redo the same work in the same way because there is a single way to do that, you do in a dedicated commit with an author that takes the clear responsibility for change. Instead my work (or a totally, byte-for-byte equivalent, one) got squashed into gigantic commits that include several very large commits of several authors (all mentioned in the commit message... but me). > To make this copyright issue somewhat relevant to GCC. GCC doesn't > currently contain individual copyright statements and most of the code > is currently assigned to the FSF. So the above mistake won't happen > when contributing to GCC, but mostly because of the technicality that > you sign away your copyright up front. Oh sorry, I wasn't clear enough about this. I'm SURE that this specific issue would not happen on GCC. Nor on Linux. Nor in several other Free Software and Open Source communities. But I think you are missing the valuable lesson that the Harvey team (some of which actually signed the rms-open-letter) tauht me: I didn't expected ANYTHING like this to happen. And I didn't expect SFC to not expell a project doing something like this. I trusted them both. All of them. So ultimately I do not expect this specific issue to occur in a hypothetical GCC lead by a Stallman-less Steering Comittee. But I DO expect that, in the long run, a Stallman-less Steering Comittee might do something not aligned with the long-term interests of Free Software, abusing my trust again. Maybe not you. Maybe not the CURRENT Steering Committee. But people, groups and incentives changes. Stallman does not. Giacomo
