Re: [PATCH 2/3] sha1_name: get_sha1_with_context learns to follow symlinks

Perhaps something like the attached patch is a good idea (I used
"BUG:" for now), I wonder?

 Documentation/CodingGuidelines | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/Documentation/CodingGuidelines b/Documentation/CodingGuidelines
index c6e536f..020370e 100644
--- a/Documentation/CodingGuidelines
+++ b/Documentation/CodingGuidelines

@@ -232,7 +232,12 @@ For C programs:
 	   to be translated, that follows immediately after it */
 	_("Here is a translatable string explained by the above.");
 
+ - Speaking of translations, do not mark the message to die() that
+   should never trigger unless there is an programming error.  E.g.
+
+	die("BUG: frotz() called with NULL pointer to xyzzy parameter");
+
- - Double negation is often harder to understand than no negation
+ - Double negation is not easier to understand than no negation
    at all.
 
  - There are two schools of thought when it comes to comparison,

On Fri, May 08, 2015 at 01:25:45PM -0700, Junio C Hamano wrote:

Perhaps something like the attached patch is a good idea (I used
"BUG:" for now), I wonder?

 Documentation/CodingGuidelines | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

Yeah, this looks good (and we definitely prefer BUG:, though I am
responsible for over 1/3 of the existing incidences).

I have been tempted to provide a macro or function for it, and have it
actually call abort() to trigger a coredump. I.e., basically assert(),
except unconditional but you get to write a more useful message.

So all of the sites would just become:

  BUG("frotz() called with NULL pointer to xyzzy parameter");

or similar.

-Peff