Pseudonymous commits

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Git gives attribution by a "name" and an "email" - however several contributors I work with are uncomfortable giving that information. I can easily use a pseudonym (I do so for myself), but is there any way to not have an email? Or shall I just use "[email protected]"?

While on the subject, I'm curious why name+email was decided upon for attribution, instead of name + optional-email + optional-URL or something that might provide additional information or useful metadata.

Thanks,
- -Mike

PS, I still have a question about git-svn & splitting repos with filter-branch under subject "dcommit-ing from a split repo" -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkuq3zQACgkQst0AR/DaKHsDZQCgxitnecj1I6NMDIindJuQna/u AoUAoNlYAQKHYx5PGYNY3MaeiKXh49qa
=etTo
-----END PGP SIGNATURE-----

Most projects want some kind of way of communicator with their contributors; and some kind of accountability with their contributors. Otherwise, how do you know whether said contributor isn't a Chinese intelligence agent trying to insert a backdoor into your program by submitting change using techniques demonstrated by the Underhanded C contest[1]? :-)

[1] http://underhanded.xcott.com/

Of course, someone can easily claim any random name, and it's not hard to get a mail account; you could pick a random name like "Mike Lifeguard", and get a gmail account, for example. :-) On the Internet, no one knows whether you are a dog. (Or a Chinese secret agent. :-)

But if someone isn't willing to give even an e-mail address, I would think even the most lax project would probably want to think twice about whether to accept patches from this contributor....

-- Ted

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Because they are only contributing translations.

Not everyone has important projects like the kernel :)

- -Mike
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkuq6YAACgkQst0AR/DaKHsMcQCgnm5/RfxlTNC0+0Mrw2jtgjez IgsAn117HtEZyf1XX88HR9vbSxZiTkQI
=1Kx1
-----END PGP SIGNATURE-----

Accountability for translations is even more important because review is usually a LOT harder.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I didn't say the translations weren't reviewed, or the translators were unaccountable. They are in both cases.

In any case, is the answer "git can't do that" or not? Because that's actually the answer I'm interested in.

Thanks,
- -Mike
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkuq7fcACgkQst0AR/DaKHvjLwCgxLt0Nf7a6u1s15hgtqOpAWAm mRcAoKnax9/54hbNXpGHlvMYRrwTj85C
=4ele
-----END PGP SIGNATURE-----

I think the point is that if you don't even have contact information for them, they can't *really* be accountable. But of course that's up to you.

Every committer (and author) in git has a (nonempty, I think) name and an email address, and that's how you identify the committer (and author). What you do with those fields is between you and your project maintainer.

Have fun,

Avery

git commit --author "I am not Mike Lifeguard <[email protected]>"

Git uses the author/committer info for diferent things. First, it is a unique identifier (name+email); second it is used when dealing with mails (in and out mails).

If you don´t use email (for review, or contributions) you can build the identifier as you want without the need to be name and email, it is up to the project. For my private projects I use name=Santi email=santi, so the author/committer info is "Santi <santi>" and I haven´t found problems. Note that I don´t send/reveive emails to contribute, just fetch and push. Maybe it would be better to use a user.email that resemble an email, with @ and such.

HTH,
Santi

Just take over authorship of the patch yourself then.

Or pick a generic email address such as the translator's mailing list email address or the like.

Nicolas

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

In the end, I managed to convince them they wouldn't be spammed (hope I'm right), and I used their real emails. Nonetheless, I appreciate the help.

If you're still in the helping mood (and know what you're talking about), I'm still looking for help with a case where an SVN repo is split into separate git repositories & being unable to dcommit changes back to SVN. The archived email is:
http://article.gmane.org/gmane.comp.version-control.git/142654

Thanks,
- -Mike
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkuruNwACgkQst0AR/DaKHsN/QCg2wyWwps6FEibNlBI5hlf9Fwd ODcAoKkcjpxNkPrVxlT690F2znoAUzlU
=rkB8
-----END PGP SIGNATURE-----

Oh, if your project is public and accessible through gitweb then spammers will certainly scan it and pick up their email addresses. Maybe some of them might even go as far as cloning Git repositories just for that purpose. Only local spam filtering may help in that case.

I know a little about Git, but not git-svn. Someone else will have to help you with that one.

Nicolas

Sounds like it'd be a really good option to have in gitweb to obfuscate email addresses a-la gmane and practically every other web based mailing list/news reader. Could be fairly low hanging fruit with a decent payoff (not that I'm volunteering). Can't do much about clone and spam.

In the case of translations one cal use generic "language team" mailing list, i.e. <[email protected]> (where LL is ISO code of language, e.g. '[email protected]' for translations to German, etc.).

Jakub Narebski
Poland
ShadeHawk on #git