[Shootout-list] Ray tracer

Robert Seeger Robert Seeger <rhseeger@gmail.com>
Fri, 29 Apr 2005 15:25:46 -0400


------=_Part_1153_3146669.1114802746373
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

See inline comments below:=20

 > I like CRAPS but whenever I use it I have to go through all of the
> > weights and set half of them to zero because the benchmarks are
> > irrelevant for me. If I could just select from about 5 categories
> this
> > would save quite a bit of time.
>=20
> What you're asking for isn't categories, it's personalization - you'd
> like the shootout webpages not to forget how you set-up the weights.


No, we're asking for categories. Though someone mentioned that it would be=
=20
nice for the page to remember how you had it setup last time you were that,=
=20
that's not what was asked for. What was asked for was a way to easily selec=
t=20
the type(s) of tests one is interested in, and view the results for those=
=20
tests.


On 4/29/05, Isaac Gouy <igouy2@yahoo.com> wrote:
>=20
>=20
> --- Robert Seeger <rhseeger@gmail.com> wrote:
> > It's not so much "solving a problem" as it is "making the collection
> > of tests more useful". If you look through some of the past
> > discussions, you'll see it mentioned that having a wide variety of
> > tests is a good thing.
>=20
> Someone might have asserted it was a good thing - I don't recall seeing
> anything to back up the opinion.


If you're refering to the idea of categories... that's what weren't=20
discussing here. If you're refering to the idea of lots of tests being good=
,=20
thats been mentioned in a number of places, both by myself and by others.


(And we're begging the question "making the collection of tests more
> useful" for what purpose?)


I would find it more convenient and useful to be able to look at only=20
certain groupings of tests at one time, easily and conveniently. I believe=
=20
that others, both in this community and out, would also find that=20
functionality handy. That is how I am defining "useful" in this=20
circumstance.


> Those examining the test results can pay attention to the tests that
> > are applicable to them. If one is interested in math, they can look
> at
> > the math tests... if not, they can ignore those tests.
> >
> > By allowing one to pick certain groupings of tests to look at easily,
> > it makes the overall test results more useful, since it's easier and
> > faster to focus on the tests that are applicable to what one finds
> > useful in a language.
>=20
> If you were the person who created the groupings then it's easier to
> focus on ... Otherwise you have to learn and remember someone else's
> idea of what group to hide the benchmark in.


Given that possiblitiy of:
* Tests don't need to be "hidden" anywhere, as selecting a group can be "in=
=20
addition" to being able to see the names of all the tests
* Tests don't need to belong to only one group
* Group names should be description "Object Oriented Programming", "Text=20
Manipulation", etc

... I don't see how adding groups would subtract from the current usability=
=20
of the tests. In general, I see only a positive impact on the "user=20
experience".


Rob Seeger

------=_Part_1153_3146669.1114802746373
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

<div>See inline comments below: <br>
<br>
</div>
<blockquote style=3D"border-left: 1px solid rgb(204, 204, 204); margin: 0pt=
 0pt 0pt 0.8ex; padding-left: 1ex;" class=3D"gmail_quote"><span class=3D"q"=
>&gt; I like CRAPS but whenever I use it I have to go through all of the</s=
pan>
<br>
  <span class=3D"q">&gt; weights and set half of them to zero because the b=
enchmarks are</span><br>
  <span class=3D"q">&gt; irrelevant for me. If I could just select from abo=
ut 5 categories</span><br>
  <span class=3D"q">this</span><br>
  <span class=3D"q">&gt; would save quite a bit of time.</span><br>
  <span class=3D"q"></span><br>
  <span class=3D"q"></span>What you're asking for isn't categories, it's pe=
rsonalization - you'd<br>
like the shootout webpages not to forget how you set-up the weights.</block=
quote>
<div><br>
No, we're asking for categories. Though someone mentioned that it would
be nice for the page to remember how you had it setup last time you
were that, that's not what was asked for. What was asked for was a way
to easily select the type(s) of tests one is interested in, and view
the results for those tests.<br>
<br>
</div><br><div><span class=3D"gmail_quote">On 4/29/05, <b class=3D"gmail_se=
ndername">Isaac Gouy</b> &lt;<a href=3D"mailto:igouy2@yahoo.com">igouy2@yah=
oo.com</a>&gt; wrote:</span><blockquote class=3D"gmail_quote" style=3D"bord=
er-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-l=
eft: 1ex;">
<br>--- Robert Seeger &lt;<a href=3D"mailto:rhseeger@gmail.com">rhseeger@gm=
ail.com</a>&gt; wrote:<br>&gt; It's not so much &quot;solving a problem&quo=
t; as it is &quot;making the collection<br>&gt; of tests more useful&quot;.=
 If you look through some of the past
<br>&gt; discussions, you'll see it mentioned that having a wide variety of=
<br>&gt; tests is a good thing.<br><br>Someone might have asserted it was a=
 good thing - I don't recall seeing<br>anything to back up the opinion.
</blockquote><div><br>
If you're refering to the idea of categories... that's what weren't
discussing here. If you're refering to the idea of lots of tests being
good, thats been mentioned in a number of places, both by myself and by
others.<br>
<br>
</div><br><blockquote class=3D"gmail_quote" style=3D"border-left: 1px solid=
 rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">(And we=
're begging the question &quot;making the collection of tests more<br>usefu=
l&quot; for what purpose?)
</blockquote><div><br>
I would find it more convenient and useful to be able to look at only
certain groupings of tests at one time, easily and conveniently. I
believe that others, both in this community and out, would also find
that functionality handy. That is how I am defining &quot;useful&quot; in t=
his
circumstance.<br>
<br>
</div><br><blockquote class=3D"gmail_quote" style=3D"border-left: 1px solid=
 rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">&gt; Th=
ose examining the test results can pay attention to the tests that<br>&gt; =
are applicable to them. If one is interested in math, they can look
<br>at<br>&gt; the math tests... if not, they can ignore those tests.<br>&g=
t;<br>&gt; By allowing one to pick certain groupings of tests to look at ea=
sily,<br>&gt; it makes the overall test results more useful, since it's eas=
ier and
<br>&gt; faster to focus on the tests that are applicable to what one finds=
<br>&gt; useful in a language.<br><br>If you were the person who created th=
e groupings then it's easier to<br>focus on ... Otherwise you have to learn=
 and remember someone else's
<br>idea of what group to hide the benchmark in.</blockquote><div><br>
Given that possiblitiy of:<br>
* Tests don't need to be &quot;hidden&quot; anywhere, as selecting a group =
can be
&quot;in addition&quot; to being able to see the names of all the tests<br>
* Tests don't need to belong to only one group<br>
* Group names should be description &quot;Object Oriented Programming&quot;=
, &quot;Text Manipulation&quot;, etc<br>
<br>
... I don't see how adding groups would subtract from the current
usability of the tests. In general, I see only a positive impact on the
&quot;user experience&quot;.<br>
&nbsp;<br>
<br>
Rob Seeger<br>
</div></div><br>
------=_Part_1153_3146669.1114802746373--