Misattributed authorship in commit log here on Github #7512

If you look here (current page for earliest commits in Bitcoin),

https://github.com/bitcoin/bitcoin/commits/master?page=289

you can see that commits done by sirius-m are attributed to Greg Maxwell.

This does not seem to be a problem with the import from SVN, as the committers show up correctly in a 'git log'. This also matches the info in the old SVN (still available on sourceforge).

Rather, it seems to be an issue or a misconfiguration on github.

I didn't check whether there are more issues like this.

The only way to solve this is to contact github. The problem is that anyone can claim ownership of any mail address. At the repository level there's no way to remove that binding. Those people are not on github, so I think this will remain a problem forever.

It's very possible that @gmaxwell claimed those addresses to prevent them from being claimed by someone else. Maybe we could set up a dummy account and move them there...

Yep, this happened while trying to figure how "saracen" was being attributed all s_nakamoto commits and reproducing them. It was reported to github and discussed in the #bitcoin-dev logs a while back. It looks like github fixed the one set and not the other.

19:29 < gmaxwell> looks like github may be compromised or badly broken: https://github.com/bitcoin/bitcoin/commits/master?author=saracen 19:34 < midnightmagic> i thought that was an artifact of the conversion process 19:37 < phantomcircuit> gmaxwell, it's always been like that 19:37 < phantomcircuit> (i had the same thought to once) 19:38 < midnightmagic> it's consistent with my (extensive) experience with scm->scm conversion tools that were badly-written 19:39 < gmaxwell> hard to see how thats possible, since that account is much newer than the commits in question. 19:39 < phantomcircuit> gmaxwell, timestamps on commits are not verified by github 19:39 < CodeShark> the commits don't have to have the right time 19:39 < CodeShark> you can set your clock back a decade and commit 19:40 < gmaxwell> thats not my point. I mean it couldn't have always been that way because that account didn't exist always. :) 19:40 < midnightmagic> argh does github order by date or by git log order..? 19:40 < CodeShark> by date - it sucks 19:43 < midnightmagic> whoah!! 19:43 < gmaxwell> yea, okay. I reproduced the stupidity. 19:45 < gmaxwell> https://github.com/bitcoin/bitcoin/commits/master?author=gmaxwell&page=6 < see bottom 19:46 < gmaxwell> first commit in bitcoin core repo is now from me. 20:13 < Luke-Jr> [02:37:19] <phantomcircuit> gmaxwell, it's always been like that <-- +1 20:14 < gmaxwell> Thats not actually true. 20:14 < gmaxwell> Unless you think it was also always like this: https://github.com/bitcoin/bitcoin/commits/master?page=264 20:14 < Luke-Jr> I don't remember a time where the contributors page didn't bug that way 20:15 < Luke-Jr> gmaxwell: it was, but with a different user 20:15 < Luke-Jr> (disclaimer: I did not view that exact page before) 20:16 < gmaxwell> perhaps we should change the merge script to refuse to merge commits that don't have dots in their email addresses. 20:16 < Luke-Jr> do we have any recent like that?
20:16 < gmaxwell> Luke-Jr: it was messed up like that from whatever moment saracen added sirius-m@1a98c847-1fd6-4fd8-948a-caf3550aa51b to their email list. 20:16 < gmaxwell> Thus my latest example, I just went and reproduced using sirius-m@1a98c847-1fd6-4fd8-948a-caf3550aa51b 20:17 < Luke-Jr> gmaxwell: sure, but that was a long time ago 20:17 < gmaxwell> oops I ment s_nakamoto@1a98c847-1fd6-4fd8-948a-caf3550aa51b in the penultimate case. 20:18 < Luke-Jr> I noticed it at least as early as when we were contacting devs to sign that joint letter
20:18 < gmaxwell> in any case, I went and reserved all the other dotless names in the history. .. looks like it only lets a single github user claim them, first come first serve.

Obviously it doesn't change the content of git itself, just what the text on the website shows.

I'll see if I can get github to fix the other ones and make them non-usurp-able like they did the first one I reported.

Cheers.

It looks like github fixed the one set and not the other.

I only reported the s_nakamoto one. I'm not sure how they resolved it. But it is a losing battle if anyone that wants to troll can just go re-claim them.

In any case, I went and reserved all the other dotless names in the history. .. looks like it only lets a single github user claim them, first come first serve.

Makes sense until a better solution can be found for them.

A better solution might be to create a github user like 'Bitcoin Unknown Author' and attribute all commits resp. email addresses that are not from the team on github to him.

'Non-Github Bitcoin author' would be a more suitable name, since some like Martti Malmi are known.

What did github assign s_nakamoto to? or is it free to claim again?

it's fixed, github has now made them nonassignable. (please close)

Awesome.

There is more:

https://archive.is/4KW50 https://archive.is/SQs6o

This is all I could find in the early history.

I have no clue-- I'll have to re-nag github; unless it's just caching. None of them are configured in my account. (The only two configured there are my gmail and xiph.org email addresses)

Re-opening this issue as it was messed up again recently.

What is the problem this time?

Same as before, old Satoshi commits are misattributed to saracen: https://github.com/bitcoin/bitcoin/commit/f1e1fb4bdef878c8fc1564fa418d44e7541a7e83

I've written Github twice about this, but it's taking long to fix this time.

The first time I did this I remember notifying somebody. I can't find the chat transcript now, I thought it happened in the bitcoin dev IRC channel, but maybe not? I also reported the problem to Github, although it may have been specifically to a developer I know that works there.

At some point, the problem appeared fixed, I was no longer the author. A few weeks/months ago I was able to attribute myself again as the author, using exactly the same method as the first time.

What you want to do about this is up to you. You could create a new account, or assign it to one of your own accounts. Or wait for Github to fix the issue.

I think github fixed it. They attribute all commits with an invalid email address to https://github.com/invalid-email-address

On Sun, Apr 2, 2017 at 4:05 PM, Arran Walker notifications@github.com wrote:

The first time I did this I remember notifying somebody. I can't find the chat transcript now, I thought it happened in the bitcoin dev IRC channel, but maybe not? I also reported the problem to Github, although it may have been specifically to a developer I know that works there.

At some point, the problem appeared fixed, I was no longer the author. A few weeks/months ago I was able to attribute myself again as the author, using exactly the same method as the first time.

What you want to do about this is up to you. You could create a new account, or assign it to one of your own accounts. Or wait for Github to fix the issue.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/bitcoin/bitcoin/issues/7512#issuecomment-290988466, or mute the thread https://github.com/notifications/unsubscribe-auth/AGGmv5RJDgzcA_WbUCWbB_sehz7JydPKks5rr6srgaJpZM4HX-Ed .

@MarcoFalke It's still an issue: https://github.com/bitcoin/bitcoin/commits?author=saracen

@MarcoFalke No, they didn't. The earliest commits made by sirius-m are still misattributed: https://github.com/bitcoin/bitcoin/commits/master?after=6d97df0e992dc5015b5327935ab89923d2dc0ba9+0

It seems that they have only marked satoshi's commits as invalid-email-address, not sirius's

Maybe sirius should claim his email so there is proper attribution? He has an account on github: https://github.com/mmalmi

The invalid-email-address account is pretty interesting. Has Github just attributed it to this account in the same way I have? I guess that's a manual process, rather than them automatically assigning invalid email addresses.

https://github.com/bitcoin/bitcoin/commits?author=invalid-email-address

Contacting support and asking them to assign it to that account is the "fix" I suppose. It might be worth finding commits with other addresses too.

@saracen I made @non-github-bitcoin so that we can attribute those commits to that account. Could you please unassociate those email addresses from your account so that I can add them to the dummy account?

I also contacted github support to see if they can/will do anything about it.

@achow101 I'm not really sure that fixes the issue.

I'll wait for a member of the bitcoin repo to reply here or for Github to assign it to the invalid-email-address account.

The invalid-email-address account is pretty interesting. Has Github just attributed it to this account in the same way I have?

I think they did. It's not something we can do ourselves at least, so would need to contact them with a list. It seems like a nice solution from them though.

Github allowed me to add all three of the emails (s_nakamoto, sirius-m, laszloh) to the @non-github-bitcoin dummy account so now those commits are properly attributed. Control of the account has been handed over to @laanwj. This issue can now be closed.

It's a shame this is Github's solution, I don't think it's ideal.

Now I just need to add those emails to another repository and then message Github in a month's time telling them they've been assigned to the wrong user "non-github-bitcoin", and whether they'd be able to reassign them for me.

Maybe they should only link to an account for verified email addresses. Anyway, I guess that's beyond the scope of the issue here :).

The real shame is that people abuse the attribution mechanism in this way, no matter whether it is possible or not.

Maybe they should only link to an account for verified email addresses. Anyway, I guess that's beyond the scope of the issue here :).

This seems the simplest solution. On the other hand, what if you committed something using an old email that you no longer have (e.g. at a company where you used to work).

Github allowed me to add all three of the emails (s_nakamoto, sirius-m, laszloh) to the @non-github-bitcoin dummy account so now those commits are properly attributed.

Thanks!

The real shame is that people abuse the attribution mechanism in this way, no matter whether it is possible or not.

I can't help but feel this was a little bit of a dig at me :) Probably correctly so... but, for what it's worth, I did reach out to Github about this and have been up front about the issue when people have mistakenly pointed out that I might be the author.

It sucks that this was/is an issue at all. I apologise that it took up time.

I did reach out to Github about this and have been up front about the issue when people have mistakenly pointed out that I might be the author.

Right, seems mostly a miscommunication issue, I was not aware that you were communicating with Github about the same thing. I complained to them so many times, I couldn't help becoming a bit annoyed to see your name associated with the commits time after time. Anyhow let's be happy it's solved.