Re: Patches for --vnlog support

[Tim Rice]

Hi Dima and Erik,

I've had a play around with the vnlog patches and done some tidying up, both to 
help them make sense to me, and also to make them more consistent with GNU and 
Datamash code style [1].

[1] https://www.gnu.org/prep/standards/standards.html#Writing-C

The attached diff is against commit 5fd50ec. The default test suite is 
currently passing.

Aside from code style updates, here is a summary of changes to Dima's original 
patches:

* We do not let vnlog claim the -v option. I agree with Erik that it should be 
reserved.

* We do not use elaborate boolean toggles to determine which 
"line_record_fread" to use. Instead, we simply add an extra argument to that 
function for switching vnlog prologue processing on or off.

* We do not adjust tests/datamash-tests-2.pl since the tests in there have 
nothing to do with vnlog. The default behavior of GNU Datamash shall not change 
for vnlog, so there is no need for those tests to change. If we agree to change 
comment handling in GNU Datamash at some point in the future, then it will 
become appropriate to update tests for comments at that time.


This isn't the last word. I will allow time for comments from Dima, Erik, and 
anyone else who is interested. The following items should also be done before 
merging:

* Address some outstanding syntax-check failures.
* Write a NEWS item.
* Ensure the style of tests/datamash-vnlog.pl matches the other test scripts.
* Ensure there is a test for each of the errors in datamash.c lines 1425-1450.
* (Optional) Add some data with vnlog formatting to the examples/ directory.

But for starters, hopefully the attached patch will make it clearer exactly 
what the change will entail for GNU Datamash.

~ Tim

vnlog.diff
Description: Text document