bug#13837: 24.2; Make it possible to turn whitespace-mode only when there are no existing problems

[Reuben Thomas]

On 27 January 2014 01:52, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

I could be convinced to install it if it's "obviously safe".
The patch as sent is a large chunk of "new" code plus a large chunk of
"removed" code, so it's not very appealing.  Diffing while ignoring
whitespace indicates it's more of a "refactoring", and by introducing an
"artificial" new function while helps keep the text lines in the
original order (and hence helps reduce the size of the diff), I get the
patch below.

Thanks very much for trying.

 

But it's not obviously safe to me.  Two non-obvious parts are:
- the removal of "(add-to-list 'whitespace-style (car option))".

The removal looks wrong to me too: without that code, the force argument has no effect. On the other hand, the way it's implemented also seems wrong, as it permanently changes whitespace-style.

 

- the change from has-bogus to bogus-list, where bogus-list will
  (initially) only be nil if whitespace-report-list is nil.

As far as I can see, has-bogus was set whenever an element was added to bogus-list. Hence, bogus-list is nil when has-bogus was nil. bogus-list is a list of whichever elements of whitespace-report-list problems are found, so if no problem is found, it will be nil.

 

Another problem is that the docstring of whitespace-test-region does not
accurately describe its return value.

So the return value is currently a list of flags corresponding to elements of whitespace-report-list.

Two more things:

1. I would like to refactor the description of whitespace problems in whitespace-report-region to move those which are the same regardless of the value of indent-tabs-mode into a separate list.

2. Since we're no longer going for getting this into 24.4, can we fold your "artificial" whitespace--report-report-region back into whitespace-report-region, which both simplifies the end result slightly, and avoids the question of whether the new function should be moved down into "internal functions"?