Re: check-news target fails if NEWS file starts with a copyright header

[Jim Meyering]

Peter Simons wrote:
> Jim Meyering wrote:
>  > The penalty for an inadequate test would be to release without updating
>  > NEWS.
>
> so, according to your logic, expecting any of the top 10 lines of NEWS to
> match a particular regular expression is "adequate", but expecting any of
> the top, say 11 lines, of NEWS to match that expression is "inadequate"?
>
> This reasoning doesn't make much sense to me. I think it's pretty obvious
> that the choice of the 10 line limit is perfectly random in this context.
> That number is inherently meaningless. Any other number, say 17 or 32, would
> work just as well for all practical purposes. My impression is that your
> opinion is based on irrational fears ("if we change the code, bad things
> will happen") rather than on rational thought.

My wanting to minimize risk is now "irrational fears"?
If you'd ever been bitten by a false-positive match (I have) and
nearly released with out-of-date NEWS, you wouldn't call it irrational.

>From your tone, I wonder if you read this message:

  http://thread.gmane.org/gmane.comp.lib.gnulib.bugs/22023/focus=22035

If it were just for my projects (as it was for years), it'd match
only on line 3, because all of my NEWS files are consistent that way.
The choice of 10 is somewhat arbitrary, but I don't want to use 15 or 20.
10 is large enough to accommodate every project that uses it so far,
while searching only 7 additional lines of change descriptions for most
(each of which represents a small risk of a false-positive match --
it is not uncommon to mention "fixed in version M.N"), but typically
no more than 6 of those additional lines are non-empty, and the odds of
getting an FP match in that range seems acceptable.

Is there some reason you'd rather not move the copyright notice
to the end of that file?

If you're motivated, propose a patch.
Make it search only lines 1..10 by default, but
allow that to be overridden via cfg.mk to search say,
only line 3 or lines 20..22.