Thread: Re: Time to add a Git .mailmap?

Re: Time to add a Git .mailmap?

From
Alvaro Herrera
Date:
On 2024-Oct-31, Daniel Gustafsson wrote:

> When looking at our Git tree for a recent conference presentation I happened to
> notice that we have recently gained duplicate names in the shortlog.  Not sure
> if we care enough to fix that with a .mailmap, but if we do the attached diff
> makes sure that all commits are accounted for a single committer entry.

LGTM.  I'd also add this line while at it:

Peter Eisentraut <peter@eisentraut.org> <peter_e@gmx.net>

This takes care of all the duplicate "identities" in the history AFAICT.

-- 
Álvaro Herrera        Breisgau, Deutschland  —  https://www.EnterpriseDB.com/
"No me acuerdo, pero no es cierto.  No es cierto, y si fuera cierto,
 no me acuerdo."                 (Augusto Pinochet a una corte de justicia)



Re: Time to add a Git .mailmap?

From
Peter Eisentraut
Date:
On 01.11.24 12:53, Alvaro Herrera wrote:
> On 2024-Oct-31, Daniel Gustafsson wrote:
> 
>> When looking at our Git tree for a recent conference presentation I happened to
>> notice that we have recently gained duplicate names in the shortlog.  Not sure
>> if we care enough to fix that with a .mailmap, but if we do the attached diff
>> makes sure that all commits are accounted for a single committer entry.
> 
> LGTM.  I'd also add this line while at it:
> 
> Peter Eisentraut <peter@eisentraut.org> <peter_e@gmx.net>
> 
> This takes care of all the duplicate "identities" in the history AFAICT.

I'm not sure if this is a good use of the mailmap feature.  If someone 
commits under <peter@companyfoo.com> for a while and then later as 
<peter@companybar.com>, and the mailmap maps everything to the most 
recent one, that seems kind of misleading or unfair?  The examples on 
the gitmailmap man page all indicate that this feature is to correct 
accidental variations or obvious mistakes, but not to unify everything 
to the extent that it alters the historical record.




Re: Time to add a Git .mailmap?

From
Daniel Gustafsson
Date:
> On 1 Nov 2024, at 13:53, Peter Eisentraut <peter@eisentraut.org> wrote:
>
> On 01.11.24 12:53, Alvaro Herrera wrote:
>> On 2024-Oct-31, Daniel Gustafsson wrote:
>>> When looking at our Git tree for a recent conference presentation I happened to
>>> notice that we have recently gained duplicate names in the shortlog.  Not sure
>>> if we care enough to fix that with a .mailmap, but if we do the attached diff
>>> makes sure that all commits are accounted for a single committer entry.
>> LGTM.  I'd also add this line while at it:
>> Peter Eisentraut <peter@eisentraut.org> <peter_e@gmx.net>
>> This takes care of all the duplicate "identities" in the history AFAICT.
>
> I'm not sure if this is a good use of the mailmap feature.  If someone commits under <peter@companyfoo.com> for a
whileand then later as <peter@companybar.com>, and the mailmap maps everything to the most recent one, that seems kind
ofmisleading or unfair?  The examples on the gitmailmap man page all indicate that this feature is to correct
accidentalvariations or obvious mistakes, but not to unify everything to the extent that it alters the historical
record.

I agree with this and propose to leave it at the originally proposed mailmap
contents.

--
Daniel Gustafsson




Re: Time to add a Git .mailmap?

From
Alvaro Herrera
Date:
On 2024-Nov-01, Peter Eisentraut wrote:

> > LGTM.  I'd also add this line while at it:
> > 
> > Peter Eisentraut <peter@eisentraut.org> <peter_e@gmx.net>
> > 
> > This takes care of all the duplicate "identities" in the history AFAICT.
> 
> I'm not sure if this is a good use of the mailmap feature.  If someone
> commits under <peter@companyfoo.com> for a while and then later as
> <peter@companybar.com>, and the mailmap maps everything to the most recent
> one, that seems kind of misleading or unfair?

While I would agree with this line of thinking if the situation were as
you describe, it should be obvious that it isn't; nobody here uses or
has ever used a work email as committer address[1][2].  Nevertheless,
since this argument is about _your_ personal identity not mine, I'm not
going to stand against you on it.

Therefore I +1 Daniel's original proposal with thanks, and BTW I'm not
sorry for changing my name to use the hard-won ' accent on it :-)

> The examples on the gitmailmap man page all indicate that this feature
> is to correct accidental variations or obvious mistakes, but not to
> unify everything to the extent that it alters the historical record.

I don't think these examples are normative.  There's plenty of evidence
that people look for ways to attribute contributions to individuals
rather than email-based identities.  See for example

https://gitlab.com/gitlab-org/gitlab/-/issues/14909
https://github.com/microsoft/vscode/blob/main/.mailmap
https://github.com/gradle/gradle/blob/master/.mailmap


[1] AFAIK gmx.net is a ISP-supplied address, not a work address.
[2] scrappy@hub.org and simon@2ndQuadrant.com might be
    considered work addresses, but they aren't really

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/
"E pur si muove" (Galileo Galilei)



Re: Time to add a Git .mailmap?

From
Daniel Gustafsson
Date:
> On 5 Nov 2024, at 10:33, Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:

> Therefore I +1 Daniel's original proposal with thanks, and BTW I'm not
> sorry for changing my name to use the hard-won ' accent on it :-)

Done.

--
Daniel Gustafsson