On 04/20/2015 06:51 AM, Thomas De Schampheleire wrote:
On Sun, Apr 19, 2015 at 4:38 PM, Mads Kiilerich <[email protected]> wrote:
On 04/19/2015 05:57 AM, Thomas De Schampheleire wrote:
I guess this is a place where a global configuration could make sense. :-(
It would be nice to show everything everywhere but especially in the tables
with commits that is not an option.
Currently:
- pull request author: username (full name)
- pull request reviewers: full name
- pull request commit overview: username only
- repo summary/changelog: username if user found, else name from commit header
- changeset detail: username (full name) if user found, else name from
commit header
(I haven't checked all other places yet, like notifications...)
Thanks - a review for consistency has value ... even more so if we can
establish guidelines for maintaining that consistency.
I would go for:
- pull request author: full name (username)
- pull request reviewers: full name (username)
- pull request commit overview: username only
For changeset/changelog displaying, I'm not fully sure: suppose
someone uses the same e-mail to commit under two different display
names, for example 'John Doe' and 'John Doe (scripted)'. In this case,
one would probably expect the name from the commit header to appear in
the changeset/changelog details.
But the correlation to the actual user as known in Kallithea is also
useful, so we should show that too, at least in the changeset details.
In case both the name in the commit header, and the name known to
Kallithea is the same, there would be some duplication if we show
both, though. Maybe we should show both but clearly indicate that one
is coming from the commit header and the other (if available) is the
detail from Kallithea.
Currently we always use the user entry and show the username if the
email address is known (and we allow the system to email the user - we
will never spam users / email addresses that are unknown to the system).
I think that is fine. If the user wants to commit under different names,
he should use different email addresses.
So then we'd have:
- repo summary/changelog: name from commit header
- changeset detail: both name from commit header as full name (username).
For other places that I did not identify above, 'full name (username)'
would be preferred, unless if there is limited space in which case
username could be shown alone.
What do you think of that?
Looks fine ... except that I like that we always use the user entry if
the email address is known and only fall back to the parsed full name if
the parsed email address is unknown.
Somewhat related: the username and email address will often have a trivial
mapping. I would like to get rid usernames and just use email addresses -
also for login, perhaps with a config option for a default @domainname that
always should be stripped.
I'm not really opposed to that, but it does mean more typing for the
typical user.
In which cases? We have completion for @annotation and reviewers ... and
the typical user will use a domain that will the default for the system
and thus can be left out.
Another thing is the confusion that comes from having separate first name
and last name fields. Cultures put given name and family name in different
order ... and sometimes people compensate for that in firstname/lastname,
sometimes they don't. I thus prefer to have a "full name" field with the
preferred spelling of the whole name and something like a "nick name" or
"common name" with the name the person usually goes by. (In addition to
that, there might be a "need" for having both the "real" name and the name
transcribed to a different culture.) This ends up as a completely different
problem but it might indicate that it could be relevant to have some kind of
configurable template for naming ... or a couple of templates for "short"
and "long" name.
I was planning to touch upon that subject in my previous reply, but
left it out because I thought it would lead us too far :)
Anyway, I think we should keep external authentication databases into
account: we should be able to map data from such databases into the
scheme we propose. Our LDAP database does have a separate firstname,
lastname, a full name and a common name, so this would be mappable on
your proposal. Note that the 'common name' in our database is not
following the local convention of 'Firstname Lastname' but is
'Lastname Firstname' regardless, and I cannot change it.
In that case it is the LDAP admins who made the decision. It is better
to expose their wrong decision that to try to fix it in the vcs system ;-)
/Mads
_______________________________________________
kallithea-general mailing list
[email protected]
http://lists.sfconservancy.org/mailman/listinfo/kallithea-general
