Le 01/11/2019 à 21:47, Tracey Emery a écrit :
> On Tue, Oct 29, 2019 at 07:02:49PM +0100, Stefan Sperling wrote:
>> On Tue, Oct 29, 2019 at 01:24:01PM -0400, Kurt Mosiejczuk wrote:
>>> OpenBSD cvs works on the "MAIN" branch by default. This diff moves got
>>> to using "main" by default rather than "master". As a side benefit, it
>>> also avoids using the potentially loaded term "master".
>> I am fine with this change. Apart from the negative connotations of
>> the terminology, promoting the idea that branch names are always chosen
>> by convention is good. Having a different default branch name than Git
>> helps to spread the idea that branch names aren't necessarily the same
>> everywhere.
>>
> Is this going to require a new got branch rename function? It's going to
> get confusing to go between got generated trees and git generated trees
> when managing both with got. I don't know. Maybe it doesn't matter and
> after time, everyone will get used to the change.
I do agree with Tracey , I am also wondering if this is not going to 
create some confusion
just to match cvs behaviour for the fun of "being different" (no offense 
intended)
Do we need to create volountary mismatches with usual git behaviour? It 
would create
harder sync with usual various git repositories,because it would 
probably mean overhead
in other developpement cases.
>>> I'm having trouble with the regression test test_import_requires_new_branch.
>>> I'm not seeing why it fails. I did verify it doesn't fail before my changes
>>> though. Maybe someone can point out what I broke.
>> The test assumes that the repository created by the test_init() function
>> contains a 'master' branch which will collide with the default 'master'
>> branch which is used by got import without your patch.
>> You can either modify the test to pass '-b master' to 'got import',
>> or add a 'main' branch in the Git repository at the beginning of the test,
>> or perhaps even modify test_init() to create a 'main' branch instead of
>> 'master' in all tests (which might cause additional test fallout).
>>
>> Regarding the man page changes:
>>
>>> -Fetch new upstream commits into the local repository's master branch.
>>> +Fetch new upstream commits into the local repository's main branch.
>>>   This step currently requires
>>>   .Xr git 1 :
>>>   .Pp
>>>   .Dl $ cd /var/git/src.git
>>> -.Dl $ git fetch origin master:master
>>> +.Dl $ git fetch origin main:main
> This is what I mean. I've already trained myself repeatedly with the
> master:master option. I suppose, it won't matter once gotd is done. I
> like the change, but it's going to take a lot of brain to adapt! :D
Indeed :)
>> Assuming that people use these instructions verbatim when trying to work
>> against the openbsd src.git repo from github, this command will now fail.
Hence my previous comment.
>> Should we adjust all examples, or just some of them?
>> Should the examples just keep using 'master' throughout?
>> I am not sure.
>>
>>>   Before outgoing changes on the local
>>> -.Dq master
>>> +.Dq main
>>> branch can be pushed to the remote repository, the local
>>> -.Dq master
>>> +.Dq main
>>>   branch must be rebased onto the
>>> -.Dq origin/master
>>> +.Dq origin/main
>>>   branch:
>>>   .Pp
>>> -.Dl $ got update -b origin/master
>>> -.Dl $ got rebase master
>>> +.Dl $ got update -b origin/main
>>> +.Dl $ got rebase main
>> Same here.
Stefan, you're GoT father, choice belongs to you :)

Kind regards,
Jerome