In article <yoijve7tvvis.fsf@remote1.student.chalmers.se>, bojohan+news@dd.chalmers.se (Johan Bockg$(D+)(Brd) writes:

> Your change replaced make_symbol with Fmake_symbol (and intern with
> Fintern), and make_symbol does

>     Fmake_symbol ((!NILP (Vpurify_flag)
>                    ? make_pure_string (str, len, len, 0)
>                    : make_string (str, len)));

> In the make_symbol/Fmake_symbol pair of functions, the Vpurify_flag
> check is in the former (so is not done after the change); but in the
> intern/Fintern pair it is in the latter. Isn't this the problem?

Ah! Perhaps.  But, I don't understand the reason of calling
make_pure_string always with the last arg multibyte as 0.
Richard, don't you remember anything?  It seems that this
part was lastly modified by you about 10 years ago.