Re: Release of version 4.14.2 of sharutils

On 01/06/15 15:56, Eric Blake wrote:
> On 01/02/2015 05:44 PM, Bruce Korb wrote:
>
>> OK, I've incorporated use of base_name and fixed the "const" attribute.
>
> No, you still missed a spot where you need to use base_name; and you
> still have some other portability bugs.  Read on...

Well, I did where it was pointed out.  I was lazy and went where pointed.

> Also, remember that tolower() doesn't always play nice in multibyte
> locales; you may want to decide if it is right to make this code
> locale-independent and multi-byte-safe by using gnulib's <c-ctype.h> and
> c_tolower(), although it limits the set of case conversions that will be
> performed to just ascii characters even in locales that have other case
> mappings defined.

At some point, it must also be remembered that the whole sharutils thing
is an old (make that, "ancient") hack for a world before MIME.
That feels like a bridge too far. :)

> Finally, there were several harmless warnings, such as:
>
> shar.c: In function 'format_report':
> shar.c:345:25: warning: unused parameter 'type' [-Wunused-parameter]
>   format_report(quot_id_t type, char const * fmt, char const * what)

It looks like the quot_id_t stuff is unnecessary in that function and
can be removed instead of quieted.  Likely a dinkleberry.

> unshar.c:62:41: warning: "/*" within comment [-Wcomment]
>    * Check for start of comment ("//" or "/*") and directives.
>   ^
>
> except that I don't know how you could reword that comment to still
> convey the same information without causing grief to (non-compliant)
> compilers that try to parse it as nested comments.

Indeed.  I think Doxycomments take priority here.  I could if-def it
away, but that uglifies the code.