[Shootout-list] Directions of various benchmarks
skaller
skaller@users.sourceforge.net
18 May 2005 15:30:37 +1000
On Wed, 2005-05-18 at 06:11, Einar Karttunen wrote:
> skaller <skaller@users.sourceforge.net> writes:
> > Why not? If you language is hard to read and you
> > need more comments to explain it, it deserves
> > to be penalised. IMHO. EG some special hack in C
> > which saves time and LOC, but which requires
> > in-depth understanding of the C Standard's rules
> > about sequence points DESERVES to be penalised:
> > good programmers should write that kind of C.
>
> In the short tests typically comments are 50% of lines.
> They mainly make notes about who contributed, tuned etc
> the benchmark. Sometimes they also make additional
> notes about the implementation.
>
> Most of the time these comments are nice and it would
> be a pity to penalize languages for them.
Why? Clutter is clutter, valuable comments are valuable,
either way: Comments cost.
See the Felix standard library which is littered
with 'doc strings' because they're needed by
the documentation processor to generate the
reference manual for the library -- I find it a pain
to maintain because of all these comments (but it
is cheaper than hand writing the library reference
manual and keeping it in synch with the library).
I actually agree with your point that some commentary
provinding meta-information (author), or explaining
the algorithm used, some technical artefact, or
whatever, can be useful.
But I suggest instead of allowing this stuff free --
because it is NOT free -- finding some fair
way to account for it.
For example: "Up to 10 lines of initial comments
are not counted in the LOC score". An arbitrary rule,
I made up on the fly .. any better idea?
--
John Skaller, mailto:skaller@users.sf.net
voice: 061-2-9660-0850,
snail: PO BOX 401 Glebe NSW 2037 Australia
Checkout the Felix programming language http://felix.sf.net