Git Rebase blows away GIT_AUTHOR_NAME

> On every fresh install of Ubuntu that I have used (by default, I use
> ecryptfs for my home directory, which of course has its own set of
> silly Git errors right now), when I clone from a remote repository,
> have changes local to me, and changes on the remote repository, do
> 'git pull --rebase', occasionally my GIT_AUTHOR_NAME gets set to my
> GIT_AUTHOR_EMAIL on my changes that get applied on top of the remote
> changes.

> Can you try running this in a repo that's giving you problems:
>
>  . git-sh-setup
>  git log --format=%H --author=your.name |
>  while read rev; do
>    get_author_ident_from_commit $rev
>    git format-patch -1 --stdout $rev |
>      git mailinfo /dev/null /dev/null
>  done | less

>> Some things off the top of my head: Does your GIT_AUTHOR_NAME or
>> GIT_AUTHOR_EMAIL contain any odd characters that might confuse a parser?
>
> Nah, from my .git/config
>
> [user]
>        name = JT
>        email = jt@instructure.com
>
>> Do you do anything special with setting up those environment variables
>> in your shell (e.g., in a .bashrc or .profile; those files _shouldn't_
>> be read by a non-interactive shell, but it's something to investigate)?
>
> Nope, the only place in my entire home directory where those get set
> are in my filter-branch script to fix them, which I only run after the
> problem manifests itself.
>
>> For that matter, how do you set up your identity in general (by
>> environment, or in ~/.gitconfig, or a local .git/config in each repo),
>> and what does it contain?
>
> I have a global identity in ~/.gitconfig
>
> [user]
>    name = JT
>    email = hello@jtolds.com
>
> and then in particular topic branches in their .git/config like before.
>
>> Can you try running this in a repo that's giving you problems:
>>
>>  . git-sh-setup
>>  git log --format=%H --author=your.name |
>>  while read rev; do
>>    get_author_ident_from_commit $rev
>>    git format-patch -1 --stdout $rev |
>>      git mailinfo /dev/null /dev/null
>>  done | less
>
> I don't have git-sh-setup, which seems like it should be included in
> the git-core package, but it's not. I have git-core
> 1:1.7.1-1.1ubuntu0.1 installed. Obviously this precludes
> get_author_ident_from_commit from working.
>
>> If that doesn't turn up anything, I think the next thing to try would be
>> making a script that reproduces the problem for you, and see if I can
>> reproduce it here.
>
> Alright, I'll see what I can do. Thanks.
>

> I got a copy of git-sh-setup. All of my commits (that have either
> worked or that I fixed) look like this:
>
> GIT_AUTHOR_NAME='JT'
> GIT_AUTHOR_EMAIL='jt@instructure.com'
> GIT_AUTHOR_DATE='1294756950 -0700'
> Author: jt@instructure.com
> Email: jt@instructure.com
> Subject: removing nondeterminism from test
> Date: Tue, 11 Jan 2011 07:42:30 -0700
>
> Should "Author" be my name? Could that be what's going on? I don't
> even know where that gets set. The ones that I failed to notice that
> they broke before I pushed them look like this:
>

> On Thu, Jan 13, 2011 at 6:11 PM, JT Olds <jtolds@xnet5.com> wrote:
>> I got a copy of git-sh-setup. All of my commits (that have either
>> worked or that I fixed) look like this:
>>
>> GIT_AUTHOR_NAME='JT'
>> GIT_AUTHOR_EMAIL='jt@instructure.com'
>> GIT_AUTHOR_DATE='1294756950 -0700'
>> Author: jt@instructure.com
>> Email: jt@instructure.com
>> Subject: removing nondeterminism from test
>> Date: Tue, 11 Jan 2011 07:42:30 -0700
>>
>> Should "Author" be my name? Could that be what's going on? I don't
>> even know where that gets set. The ones that I failed to notice that
>> they broke before I pushed them look like this:
>>
>
> "git am" (which git rebase builds on) requires the author name to be
> at least three characters long. This is a problem that has been
> discussed before, see
> <AANLkTinqTL7gH4CHEfy8UrhK13xcO_3UzgIyQka00MAh@mail.gmail.com>:
>
> http://mid.gmane.org/AANLkTinqTL7gH4CHEfy8UrhK13xcO_3UzgIyQka00MAh@mail.gmail.com
>

> On Thu, Jan 13, 2011 at 10:47 AM, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>> On Thu, Jan 13, 2011 at 6:11 PM, JT Olds <jtolds@xnet5.com> wrote:
>>> I got a copy of git-sh-setup. All of my commits (that have either
>>> worked or that I fixed) look like this:
>>>
>>> GIT_AUTHOR_NAME='JT'
>>> GIT_AUTHOR_EMAIL='jt@instructure.com'
>>> GIT_AUTHOR_DATE='1294756950 -0700'
>>> Author: jt@instructure.com
>>> Email: jt@instructure.com
>>> Subject: removing nondeterminism from test
>>> Date: Tue, 11 Jan 2011 07:42:30 -0700
>>>
>>> Should "Author" be my name? Could that be what's going on? I don't
>>> even know where that gets set. The ones that I failed to notice that
>>> they broke before I pushed them look like this:
>>>
>>
>> "git am" (which git rebase builds on) requires the author name to be
>> at least three characters long. This is a problem that has been
>> discussed before, see
>> <AANLkTinqTL7gH4CHEfy8UrhK13xcO_3UzgIyQka00MAh@mail.gmail.com>:
>>
>> http://mid.gmane.org/AANLkTinqTL7gH4CHEfy8UrhK13xcO_3UzgIyQka00MAh@mail.gmail.com
>>
>
> Oh thank you Erik, it's great to just know what the problem is. I've
> been feeling like some voodoo was happening.
>
> What are the chances of decreasing that lower bound of author name size? :)
>

> On Thu, Jan 13, 2011 at 19:20, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>>[..] Besides, a name of
>> two characters aren't really sane. You'd need at least three
>> characters to form a first/last name pair.
>
> I think I've mentioned this before in another thread, but first/last
> name isn't universal, not even within countries where it's the common
> form. When I was as student there was a fellow student from another
> scandinavian country and his legal, full name consisted of a single
> letter.
>

> On Fri, Jan 14, 2011 at 9:45 AM, Tor Arntsen <tor@spacetec.no> wrote:
>> I think I've mentioned this before in another thread, but first/last
>> name isn't universal, not even within countries where it's the common
>> form. When I was as student there was a fellow student from another
>> scandinavian country and his legal, full name consisted of a single
>> letter.
>>
>
> I'm curious, what Scandinavian country was this? Because as a
> Norwegian, I know a lot of people from all Scandinavian country, yet
> I've never heard of such names. In Norway, I the shortest legal name
> I've ever heard of was five characters.

> On Fri, Jan 14, 2011 at 09:56, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>> On Fri, Jan 14, 2011 at 9:45 AM, Tor Arntsen <tor@spacetec.no> wrote:
>>> I think I've mentioned this before in another thread, but first/last
>>> name isn't universal, not even within countries where it's the common
>>> form. When I was as student there was a fellow student from another
>>> scandinavian country and his legal, full name consisted of a single
>>> letter.
>>>
>>
>> I'm curious, what Scandinavian country was this? Because as a
>> Norwegian, I know a lot of people from all Scandinavian country, yet
>> I've never heard of such names. In Norway, I the shortest legal name
>> I've ever heard of was five characters.
>
> Sweden (I'm Norwegian too - this guy was a Swede studying in Norway).
> Admittedly I have only that single example, and it was back in the
> late seventies. His name was accepted as legal by Statens Lånekasse
> (bank for students) and when the loans arrived his single-letter name
> would be found at the very end of the long lists of wide listing-paper
> printouts from the bank that was stiched up on the billboard wall
> outside the administration offices. The loans arrived a couple of
> times per year but we always had to go looking - the rest of us were
> just amazed that we could really find that single letter down there
> and he wasn't bs'ing the rest of us about his name.
>
> I'm not sure why there's a 3-letter limit on git author names.. but I
> would suggest it should be set down to 1 letter minimum.. below that
> would, I think, be overdoing it..
>

> (CC'ed Linus, as he wrote mailinfo's sanity-checking)
>
> On Fri, Jan 14, 2011 at 10:24 AM, Tor Arntsen <tor@spacetec.no> wrote:
>> On Fri, Jan 14, 2011 at 09:56, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>>> On Fri, Jan 14, 2011 at 9:45 AM, Tor Arntsen <tor@spacetec.no> wrote:
>>>> I think I've mentioned this before in another thread, but first/last
>>>> name isn't universal, not even within countries where it's the common
>>>> form. When I was as student there was a fellow student from another
>>>> scandinavian country and his legal, full name consisted of a single
>>>> letter.
>>>>
>>>
>>> I'm curious, what Scandinavian country was this? Because as a
>>> Norwegian, I know a lot of people from all Scandinavian country, yet
>>> I've never heard of such names. In Norway, I the shortest legal name
>>> I've ever heard of was five characters.
>>
>> Sweden (I'm Norwegian too - this guy was a Swede studying in Norway).
>> Admittedly I have only that single example, and it was back in the
>> late seventies. His name was accepted as legal by Statens Lånekasse
>> (bank for students) and when the loans arrived his single-letter name
>> would be found at the very end of the long lists of wide listing-paper
>> printouts from the bank that was stiched up on the billboard wall
>> outside the administration offices. The loans arrived a couple of
>> times per year but we always had to go looking - the rest of us were
>> just amazed that we could really find that single letter down there
>> and he wasn't bs'ing the rest of us about his name.
>>
>> I'm not sure why there's a 3-letter limit on git author names.. but I
>> would suggest it should be set down to 1 letter minimum.. below that
>> would, I think, be overdoing it..
>>
>
> Linus, you wrote sanity_check (from 2744b23). Do you remember if there
> were any specific reason for the minimum length of 3 of an
> author-name? It seems that in Sweden, legal names can be even a single
> letter (see Tor's comment)...
>

>
> Linus, you wrote sanity_check (from 2744b23). Do you remember if there
> were any specific reason for the minimum length of 3 of an
> author-name? It seems that in Sweden, legal names can be even a single
> letter (see Tor's comment)...

> On Fri, Jan 14, 2011 at 1:53 AM, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>>
>> Linus, you wrote sanity_check (from 2744b23). Do you remember if there
>> were any specific reason for the minimum length of 3 of an
>> author-name? It seems that in Sweden, legal names can be even a single
>> letter (see Tor's comment)...
>
> Even if the legal name would be a single letter, you'd still need to
> have a surname.
>

> That thing is supposed to be a *NAME*. Not shorthand. Not your first
> name. Not your nickname. If you have a nickname, put it in quotes
> inside the real name.
>
> I've seen too many broken source control systems that just take your
> login as a name *cough*CVS*cough*, and then people think it's
> "convenient" and "cool" to have a short name.
>
> It's not convenient. It's not cool. It's just shorthand where
> shorthand doesn't help. Then you end up using it in a public setting,
> and suddenly your cool shorthand or nickname isn't even remotely
> unique.
>
> No, there is no uniquness "requirements" for the name, but come on.
> Look at shortlog output some day. We try to use just the name because
> it looks better. But if people don't use their full name, it just
> looks _stupid_

> > The three-letter minimum is just a sanity check. If your name really
> > is even just three letters, I suspect you're just lying. I don't know
> > of anybody named "A B".
> >
> Thanks for clarifying that it's not there for a technical reason. The
> thing is, git-am seems to be the only place where such a sanity-check
> is performed. Shouldn't git-commit rather perform such checks also (if
> such a check should be done at all), perhaps with an override similar
> to --allow-empty? And on top of all it doesn't barf, it just silently
> replace the name with the e-mail...

> On Fri, Jan 14, 2011 at 05:13:59PM +0100, Erik Faye-Lund wrote:
>
>> > The three-letter minimum is just a sanity check. If your name really
>> > is even just three letters, I suspect you're just lying. I don't know
>> > of anybody named "A B".
>> >
>> Thanks for clarifying that it's not there for a technical reason. The
>> thing is, git-am seems to be the only place where such a sanity-check
>> is performed. Shouldn't git-commit rather perform such checks also (if
>> such a check should be done at all), perhaps with an override similar
>> to --allow-empty? And on top of all it doesn't barf, it just silently
>> replace the name with the e-mail...
>
> I tend to agree with Linus on the stupidity issue, but I do worry about
> the subtlety of the results. It causes silent data corruption during a
> rebase (or when somebody is applying an emailed patch). On the other
> hand, I do understand why Linus made a sanity check in the first place;
> his use case is to deal with whatever crap people happen to mail him,
> whether they have used git or not.
>
> So we should probably do one or both of:
>
>  1. Make an --allow-any-name option to mailinfo, and use it when we
>     invoke mailinfo internally for rebasing. That still doesn't solve
>     the emailed patch problem, but at least keeps purely internal
>     operations sane.
>
>  2. Bump the check up to git-commit time, which is the best place to
>     catch and tell somebody that their name is too short, because they
>     can actually fix it.
>

> >  2. Bump the check up to git-commit time, which is the best place to
> >     catch and tell somebody that their name is too short, because they
> >     can actually fix it.
> >
> 
> The problem with (2) is that git-am uses git-commit-tree rather than
> git-commit. But I do think that adding the same checks to git-commit
> would make sense. Unless we decide to remove the checks, that is...

> So we should probably do one or both of:
>
>   1. Make an --allow-any-name option to mailinfo, and use it when we
>      invoke mailinfo internally for rebasing. That still doesn't solve
>      the emailed patch problem, but at least keeps purely internal
>      operations sane.
>
>   2. Bump the check up to git-commit time, which is the best place to
>      catch and tell somebody that their name is too short, because they
>      can actually fix it.
>
> Even if we dropped the check now, option (2) is still useful, because
> you have no idea which version of git the other end will use to apply
> your patch.

> Jeff King <peff@peff.net> writes:
> 
> > So we should probably do one or both of:
> >
> >   1. Make an --allow-any-name option to mailinfo, and use it when we
> >      invoke mailinfo internally for rebasing. That still doesn't solve
> >      the emailed patch problem, but at least keeps purely internal
> >      operations sane.
> >
> >   2. Bump the check up to git-commit time, which is the best place to
> >      catch and tell somebody that their name is too short, because they
> >      can actually fix it.
> >
> > Even if we dropped the check now, option (2) is still useful, because
> > you have no idea which version of git the other end will use to apply
> > your patch.
> 
> I am perfectly Ok with making the check looser in "am" when $rebasing is
> in effect.  Wouldn't that solve the issue?

> So we should probably do one or both of:
> 
>   1. Make an --allow-any-name option to mailinfo, and use it when we
>      invoke mailinfo internally for rebasing. That still doesn't solve
>      the emailed patch problem, but at least keeps purely internal
>      operations sane.

> On Fri, Jan 14, 2011 at 11:21:45AM -0500, Jeff King wrote:
>
>> So we should probably do one or both of:
>>
>>   1. Make an --allow-any-name option to mailinfo, and use it when we
>>      invoke mailinfo internally for rebasing. That still doesn't solve
>>      the emailed patch problem, but at least keeps purely internal
>>      operations sane.
>
> So I wrote up a nice tidy patch series with mailinfo changes and tests,
> and then am/rebase changes and tests on top of that. And guess what I
> noticed? My rebase tests didn't actually fail with stock git.
>
> I bisected to 43c2325 (am: use get_author_ident_from_commit instead of
> mailinfo when rebasing, 2010-06-16), which was written to handle exactly
> this sort of thing.
>

> On Fri, Jan 14, 2011 at 11:21:45AM -0500, Jeff King wrote:
>
>> So we should probably do one or both of:
>>
>>   1. Make an --allow-any-name option to mailinfo, and use it when we
>>      invoke mailinfo internally for rebasing. That still doesn't solve
>>      the emailed patch problem, but at least keeps purely internal
>>      operations sane.
>
> So I wrote up a nice tidy patch series with mailinfo changes and tests,
> and then am/rebase changes and tests on top of that. And guess what I
> noticed? My rebase tests didn't actually fail with stock git.
>
> I bisected to 43c2325 (am: use get_author_ident_from_commit instead of
> mailinfo when rebasing, 2010-06-16), which was written to handle exactly
> this sort of thing.
>
> That commit made it into v1.7.2. JT, which version of git are you
> running? Do you still see any problems with v1.7.2 and above? I can't
> replicate your issue with more recent versions.
>
> So unless we want to do any sort of commit-time warning, I don't think
> there is anything left to be done on this topic.
>
> -Peff
>

> On 14/01/2011 17:13, Erik Faye-Lund wrote:
>
>> I think Tor pointed out that he knew a swede with his full legal name
>> to be only one letter long. I would suppose that meant that he didn't
>> have a surename?
>
> Exactly. He didn't. Bank printouts etc. would only have that single
> letter, he didn't use a nickname - that letter was his legal name.
>

> On 14/01/2011 17:13, Erik Faye-Lund wrote:
>
>> I think Tor pointed out that he knew a swede with his full legal name
>> to be only one letter long. I would suppose that meant that he didn't
>> have a surename?
>
> Exactly. He didn't. Bank printouts etc. would only have that single
> letter, he didn't use a nickname - that letter was his legal name.
>
> As for the rest of the world - I don't think the first name/last name combo
> (almost) everyone in the west use is necessarily a universal rule.

> On Fri, Jan 14, 2011 at 09:56, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>> On Fri, Jan 14, 2011 at 9:45 AM, Tor Arntsen <tor@spacetec.no> wrote:
>>> I think I've mentioned this before in another thread, but first/last
>>> name isn't universal, not even within countries where it's the common
>>> form. When I was as student there was a fellow student from another
>>> scandinavian country and his legal, full name consisted of a single
>>> letter.
>>>
>>
>> I'm curious, what Scandinavian country was this? Because as a
>> Norwegian, I know a lot of people from all Scandinavian country, yet
>> I've never heard of such names. In Norway, I the shortest legal name
>> I've ever heard of was five characters.
>
> Sweden (I'm Norwegian too - this guy was a Swede studying in Norway).
> Admittedly I have only that single example, and it was back in the
> late seventies. His name was accepted as legal by Statens Lånekasse
> (bank for students) and when the loans arrived his single-letter name
> would be found at the very end of the long lists of wide listing-paper
> printouts from the bank that was stiched up on the billboard wall
> outside the administration offices. The loans arrived a couple of
> times per year but we always had to go looking - the rest of us were
> just amazed that we could really find that single letter down there
> and he wasn't bs'ing the rest of us about his name.
>
> I'm not sure why there's a 3-letter limit on git author names.. but I
> would suggest it should be set down to 1 letter minimum.. below that
> would, I think, be overdoing it..
>

> Author: jt@instructure.com
> Email: jt@instructure.com
> Subject: removing nondeterminism from test
> Date: Tue, 11 Jan 2011 07:42:30 -0700
> 
> Should "Author" be my name? Could that be what's going on? I don't
> even know where that gets set. The ones that I failed to notice that
> they broke before I pushed them look like this: