Hello,
Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Hello,
>
> pierre.techoueyres@free.fr (Pierre Téchoueyres) writes:
>
>> I think I've corrected all points. You'll find new versions attached.
>
> Thank you.
>
>> Would you mind consider to include the patch for the detection of
>> encoding with the #+include keyword in 9.2 release ?
>
> This patch is still missing some small parts for proper integration,
> namely documentation, and, if possible, a couple of tests. Besides, 9.2
> branch is supposedly frozen.

I argree for the documentation and tests (but I have to admit I don't
know how to add them).

> Granted, it doesn't seem too harmful, but is there any strong reason to
> integrate it in Org 9.2 (assuming documentation is ready)?


I think I wasn't clear enough : I had hope you will only include the
part which correct the decoding of include keyword, not the whole two
patchs. I think the former is simply a bug fixes. 


>
>> +		   (coding
>> +		    (intern (or (and (string-match
>> +
>> ":coding[[:space:]]+\\_<\\(\\(?:\\sw\\|\\$\\|&\\|\\*\\|\\+\\|-\\|_\\|<\\|>\\)+\\)\\_>"
>> value)
>> +				     (prog1 (match-string 1 value)
>> +				       (setq value (replace-match "" nil nil value))))
>> +				(symbol-name coding-system-for-read))))
>
> I suggested a refactoring that you didn't integrate: it seems wasteful
> to call `intern' on the return value of `symbol-name'.
>
> Besides, my suggestion about the regexp was wrong. We shouldn't make the
> syntax foolproof. I think
>
>   ":coding +\\(\\S-+\\)"
>
> is enough actually. Sorry about sending you in the wrong track.
>
> Regards,

Here is a new amended patch.