[Gambas-user] sourceforge , git and gitlab, time to switch to gitlab

[Adrien Prokopowicz]

Le Mon, 13 Mar 2017 15:22:48 +0100, Benoît Minisini  
<gambas at ...1...> a écrit:

> Le 13/03/2017 à 15:16, PICCORO McKAY Lenz a écrit :
>> i'm back for this land and in my firts 2017 mail to this i see some  
>> notes
>> about sourgeforce
>>
>> 2017-02-25 20:37 GMT-04:00 Jesus Guardon <jguardon at ...3606...>:
>>
>>> El 25/2/17 a las 20:17, Benoît Minisini escribió:
>>>> Moreover, the Sourceforge svn commit mails are still broken, they
>>>> apparently haven't fixed them for weeks. Sourceforge is more and more  
>>>> a
>>>> problem!
>>> Perhaps it's time to switch to GitHub...  ;-)
>>>
>>>
>> i best for gambas prefer gitlab due gitlab are more GPL rather github,  
>> the
>> github interface are copyright and gitlab are 100% GPL and the  
>> enterprice
>> services only add some services over the community..
>>
>> so when we use the gitlab ui service we follow gpl guidelines, also  
>> helps a
>> great project that share their ui code, either github not!
>>
>> * and a important difference, github projects are a sub.item of github
>> repos.. that are confuse!
>> * in gitlab, the repository its a subitem of a project, so the  
>> organization
>> are more proper!
>>
>> also gitlab offers mirror git autocommit to other repositories so we can
>> sync the older sourceforge using svn hook and mirror git, gitlab does  
>> not
>> offer automatic mirror commits, only using git remote configuration  
>> inside
>> repository.
>>
>> these 3 main reason means that gitlab are better.. and the right choice
>>
>> Lenz McKAY Gerardo (PICCORO)
>> http://qgqlochekone.blogspot.com
>
> Sebastian Kuelez prefers github: maybe he can give his arguments on the
> mailing-list?
>
> Anyway, whatever repository is chosen, the only problem I have is how to
> import the svn commit history with the user that did the commit, as
> these users won't exist in the new git repository.
>

Git doesn't actually manage users itself : for each commit, it only records
a full name and an email, both of which are purely indicative.
Therefore, it is not a problem at all that non-existent authors are  
mentioned
in commits (as they don't actually exist anywhere anyway).
It will only prevent the front-end (GitHub, GitLab, etc.) to display the  
user
icons, and it may mess up the stats a bit. But all of the history (commits  
and
branches) is good (I just tested it on a GitHub private repository).

When using git-svn[0] to import the SVN repository, you can also give an
authors file, which binds the SVN usernames to full names and emails,
hence fixing this problem, making everything nice and clean. :)

[0] https://git-scm.com/docs/git-svn

-- 
Adrien Prokopowicz