Re: Column labels with --full

[Tim Rice]

> Maybe document it as deprecated, print a warning about the
> combination, and turn it into a full blown error in a few releases?

Hey all,

After looking more closely at the issues, I've come to agree with Shawn on this 
one. Rather than trying to invent a new meaning for something that never really 
made sense, let's deprecate non-linewise uses of -f/--full.

If there were other problems like headers not having the right operations 
printed, let's consider that a separate issue. For starters, such a task would 
be a fair bit more difficult. Secondly, it will be easier to reason about the 
issues when --full is not complicating the matter.

The attached patch makes an attempt at deprecating -f/--full for all operations 
not in the MODE_PER_LINE category.

The largest amount of work was to make sure the tests still make sense. All the 
tests involving --full need to be assessed for whether they make sense. The 
ones that don't require either rejigging to use a more sensible operation, or 
else delete them.

I'm not quite sure if I did this right. Eg, when a test is deleted, should all the tests 
after it be re-numbered to remain consecutive? Or preserve the original numbering, so 
that all tests keep the same "uuid" throughout history?

Well, let me know what you think, or I'll merge it if I come back after a day 
or two and it still looks okay to me.

~ Tim

deprecate_full_groupby.patch
Description: Text document