On Sat, Apr 06, 2019 at 01:07:23AM +0100, Radostin Stoyanov wrote:
> $ cd fs/
> $ find . -type f -exec sed --in-place 's/[[:space:]]\+$//' {} \+
>
> Signed-off-by: Radostin Stoyanov <rstoyanov1@xxxxxxxxx>
No. That's not the way to do that kind of stuff. First, how it
should be done (and that really belongs somewhere in Documentation/):
after having convinced Linus that mechanical change in question needs
to happen, you ask him to run the script in question just before the
-rc1 of the next cycle.
Reason for _not_ doing it as you have: you are creating a pile of
conflicts with any number of development branches in various trees,
for no good reason. It's a bloody bad idea, especially for something
this trivial.
What's more, some of those (and the most heavily affected ones) bear
rather interesting comments:
* This translation table was generated automatically, the
* original table can be download from the Microsoft website.
* (http://www.microsoft.com/typography/unicode/unicodecp.htm)
_If_ it's autogenerated, I'd suggest leaving it alone, or modifying
whatever tool has been used to produce the damn thing. FWIW, the
situation might be even more unpleasant - URL in the comment is
stale. Finding the source actually used is not trivial -
poking on archive.org gives multiple versions of the data
that comment probably refers to.
It would be nice to straighten that mess - as it is, we are
probably not stepping into the section 3 there, but "the arrays
below had been produced by some tool (not included) from some
version(s) of the tables once reachable via links on now-defunct
webpage at $URL" is not a good situation, even if nobody really
gives a damn about those tables...
