I think it would be very nice to include Real Names in MAINTAINER
lines, just as OpenBSD does. Stephen@ floated this idea originally, so credit
should go to him.
There is no technical reason not to do this, and the patch attached will keep the old behaviour for make describe and make maintainer. The only code that could suffer is code that runs make -VMAINTAINER rather than make maintainer.
Matthew, p5-FreeBSD-Portindex runs make describe, so no problems there.
Shaun, portscout will run make -VMAINTAINER if it can't find the maintainer address in INDEX. I can't see this ever happening... but you can replace -VMAINTAINER with -V_MAINTAINER and it will work just fine.
Dan, how does freshports get its information? Will it survive with maintainer entries like that?
Joe, portlint should stop complaining about this .
How does pointyhat get the MAINTAINER line, and again will it care about the extra data if it doesn't use make maintainer ? I can't find the source code, so I can't check it.
I'm not aware of any issues with this... I'm sure if anything does break, the simplest solution is to replace MAINTAINER with _MAINTAINER in the offending code.
The autoassigner will still work, because it's still a valid email address.
Fix: I think that the best way to go about this would be to strictly limit the changes initally to a small subset of ports, and iron out problems if they appear. It's really not a big change.
bsd.port.mk is portmgr territory (via the GNATS Auto Assign Tool)
Initial tests show show FreshPorts is OK with that.
See http://dev.freshports.org/sysutils/bacula-server/ where I faked a =
FreshPorts gets nearly everything via make -V, and in this case, make -V =
The expected results wind up in the database:
freshports.org=3D# select maintainer from ports where id =3D (select id =
from ports_active where name =3D 'bacula-server');
Dan Langille <firstname.lastname@example.org>
Any problems that arise won't be database related (it clearly captures =
the right stuff). Therefore, +1 from me.
We might have to amend searches, or perhaps alter the opt-in sequence =
for committers, but I can look into that later. For now,
such a change would not break incoming commits. They will still be =
processed, and logged into the database.
Dan Langille - http://langille.org
I'll take it.
Actually p5-FreeBSD-Portindex does not run 'make describe' -- it
effectively runs 'make -V MAINTAINER' to extract the maintainer email
Will need a small patch to the Portindex code to allow it to reproduce
the correct INDEX format, but don't let that prevent this change being
Dr Matthew J Seaman MA, D.Phil.
Is this PR still relevant?
Yes, still relevant :)
Given how many scripts there are in the wild, we should keep make -VMAINTAINER presenting the mail only part. (which does not prevent from using MAINTAINER= <fullname> <mail> in the Makefile itself.
and probably propose a MAINTAINER_FULL variable or something like that will the full name.
That said if we do start going in that direction we should also defined a preferred encoding as name comes with encoding troubles in many countries