Comment 1 for bug 1348614

Thanks. Fixes for all three committed, although #3 is perfectly clear use of the ternary operator IMO (but terrible for “division with rounding up”, so I replaced the code).

I’m developing on MirBSD against GCC 3.4 (heavily patched), so I appreciate reports from newer/other compilers and static or even dynamic analysers. (I do occasionally try clang scan-build on my code, and build on e.g. GNU/Linux with recent GCC, but not continuously.)

I dropped the comparison in #1 entirely (probably historic) and changed the loop in #2 to use mksh’s NELEM macro, since the size of the speeds[] array is compile-time dynamic and dependent on the host OS.

Feel free to re-check CVS HEAD (or the inofficial mirror on github).