• add ${PY_FLAVOR} to python dependency
• fix distinfo

Fix error with i18n ports identified by exp-run. I think this is ready to ship :-)

I've opted for your second suggest: break dependency of the language ports on eric6 and instead mention them in a pkg-message.

Yay, ancillary ports have been committed. I'm slowly flushing out the backlog of changes that I've accumulated.

Give full context (apologies, I keep forgetting) and order the USES block.

In D12440#286673, @russ.haley_gmail.com wrote:

Hi David,

In the end, this review is just too massive for me to work with. I had some issues with the patch for mono but was able to get it to build. I'd like to help and submit that patch, but I can't work with all the other changes too. Can I suggest breaking this review up along the groupings in your comment and submitting each part separately (starting with the mono update)? I just don't think this update will ever get finished due to it's breadth.

Enhance ignoring of svn-commit*.tmp and align ignored work directories between git and subversion.

Thanks for the review @mat

P.S. Phabricator isn't doing a great job of showing the changes to the subversion attributes. The unified diff looks like this:

--- .   (revision 455899)
+++ .   (working copy)

Apologies, it looks like I didn't update the patch :-(.

Is there any other change suggested/required? Can this change be committed (with flavors)?

In D10698#221845, @adamw wrote:

Also, consider adding /*/*/Makefile.local.

Patch rejected: see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=198881 and https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=208083

• Add a default FLAVOR, after defining FLAVORS.
• Remove redundant ".include <bsd.port.pre.mk>"

Fixed in r456293

Thanks, I has misunderstanding when FLAVOR gets defined (I thought only after including bsd.port.options.mk, or co). Fixed that and removed redundant code.

In D13449#281454, @mat wrote:

I prefer it the way it currently is, it shows in a very simple way where PY_FLAVOR come from.

In D13449#281299, @mat wrote:

Ok, I see what you trying to do.

It is by design that the selection of the Python version through flavors is not limited to when python.mk sets FLAVORS, so this is wrong. I will try to think of something for people who use conflicting flavor names.

In D13449#281338, @mat wrote:

In D13449#281137, @dbn wrote:

Note that the above change only fixes UNIQUE_DEFAULT_LINKS if python flavors is used. In the following example it does not work:

# cd /usr/ports/Mk
# make -f bsd.port.mk -V UNIQUE_DEFAULT_LINKS USES=python:3.3+ USE_PYTHON=concurrent
no
# make -f bsd.port.mk -V UNIQUE_DEFAULT_LINKS USES=python:3.3+ USE_PYTHON="concurrent flavors"
yes

Is there actually one port that does this ? I was trying to test the DEV_WARNING I was adding, and I could not find any.

Update patch per comments.

In D13448#281293, @mat wrote:

It would probably be easier to have the flavors named qtX_pyY, so you don't conflict with the flavors from USES=python.

I assume you are only referring to point 3, as the other two points were not addressed in commit r456026. I think we also disagree on the definition of "really, really, really simpler":

I would suggest you enhance qa.sh to check for shebangs with python2 and python3 (since this depreciates a port depending on a python metaport).

Rebase to r453757

Changes:

• rebase patch to r450479
• lang/mono: fix distinfo with ACCEPTANCE_TESTS
• spelling fix in summary
• update log in summary