Re: trimming getloadavg's dependencies

[Jim Meyering]

Paul Eggert wrote:
> When I tried to add the getloadavg module to GNU Emacs,
> I brought in a whole bunch of new files (see below)
> due to all the dependencies.  This seems like way overkill.

Wow.  That does seem excessive.

> For example, if the c-strtod module is not otherwise being
> used, we can break the dependency on c-strtod by
> parsing the floating point number by hand; that may not
> be as precise but it's good enough for a load average.
>
> Any objections to trying that sort of thing?

Please do.  There is so little precision in load average
numbers (and probably never an exponent), that nothing would be lost.

> The list below really does seem excessive.
>
>   lib/c-strtod.c
>   lib/c-strtod.h
>   lib/cloexec.c
>   lib/cloexec.h
>   lib/creat-safer.c
>   lib/dup-safer.c
>   lib/dup2.c
>   lib/fcntl--.h
>   lib/fcntl-safer.h
>   lib/fcntl.c
>   lib/fcntl.in.h
>   lib/fd-safer.c
>   lib/getdtablesize.c
>   lib/locale.in.h
>   lib/malloc.c
>   lib/open-safer.c
>   lib/open.c
>   lib/pipe-safer.c
>   lib/stat.c
>   lib/stdint.in.h
>   lib/stdlib.in.h
>   lib/strdup.c
>   lib/string.in.h
>   lib/sys_stat.in.h
>   lib/unistd--.h
>   lib/unistd-safer.h
>   lib/wchar.in.h
>   m4/cloexec.m4
>   m4/dos.m4
>   m4/dup2.m4
>   m4/fcntl-o.m4
>   m4/fcntl-safer.m4
>   m4/fcntl.m4
>   m4/fcntl_h.m4
>   m4/getdtablesize.m4
>   m4/locale_h.m4
>   m4/longlong.m4
>   m4/malloc.m4
>   m4/mode_t.m4
>   m4/open.m4
>   m4/stat.m4
>   m4/stdint.m4
>   m4/stdlib_h.m4
>   m4/strdup.m4
>   m4/string_h.m4
>   m4/sys_stat_h.m4
>   m4/unistd-safer.m4
>   m4/wchar_h.m4
>   m4/wint_t.m4