Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: Discussion paper: what are the level 2 datums?



John, P1788,

In my opinion, there is a large degree of compatibility and overlap between the proposed NaI and a bare decoration. In other words, the NaI is mostly serving the same purpose as a bare decoration. This shouldn't come as too much suprise, because it was the intention of Motion 8 that a "bare decoration" is NaI.

But the paper is also suggesting a few changes that will open doors that I see can be very dangerous, and which the Motion 8 concept of a bare decoration specifically was intended to prevent. So I STRONGLY urge you and P1788 not to abolish bare decorations!

I would instead suggest the motion should be carefully modified so as to merge the NaI concept with that of a bare decoration more cleanly. This was the original intention of Motion 8, anyways, and I think this current paper could go a long way to making such a merger more concise and specific (with a little care, I think this could be done).

Let me just back up a little and give a background and summary:

One of the main purposes of bare decorations is to guarantee that it _always_ propagates through a lengthy computation. In this respect it is not too different than what you are proposing. However, your proposal opens the door that some NaI could be "absorbed" in a lengthy computation since you suggest that there is an operation:
   bare(NaI) = Empty
which in my opinion is VERY dangerous.

In motion 8, there are:
   decorated intervals
   bare intervals
   bare decorations (NaI)
a decorated interval is, of course, a composite object consisting of both a bare interval and a bare decoration. The idea is one can always strip away either the interval or decoration portion of a decorated interval. However, it is nonsense to suggest that one can "strip away" the interval portion of a bare decoration (NaI) or the decoration portion of a bare interval.

However, it is appropriate that a bare interval could be "promoted" to a decorated interval, e.g., there could be some decorated interval constructor that accepts as an argument only a bare interval, and the decorated interval will be initialized appropriately (see, e.g., clause 2.4 of motion 8).

But motion 8 was specifically designed so that a bare decoration (NaI) could never be "promoted" or "converted" into either a bare interval or a decorated interval (see, e.g., clause 2.3). In that regard, Motion 8 was intentionally designed to NOT allow some construct like:
   bare(NaI) = Empty
specifically for the reasons mentioned above.

Actually, if you dig up the archives and look at the first version of motion 8, you will see that at that point in time we were suggesting the same thing you are now suggesting: that the bare interval portion of a bare decoration was an "implied" empty set.

However, it was Arnolds observation that this was dangerous, for the reasons mentioned above. I also agree with this very much. So this led to a revision of Motion 8 before voting began, which is the 8.02 version on the IEEE website (which introduced clause 2.3 as it was then voted on).

So I think that is the one big problem with the current paper. With a little modification, though, I think this could be reversed. Most of the other concepts in the paper I think are heading in the right direction.

Of course, you are the tech writer, so I think it is good you are trying to find ways to explain all this in your own words.

Sincerely,

Nate Hayes






----- Original Message ----- From: "John Pryce" <j.d.pryce@xxxxxxxxxxxx>
To: "stds-1788" <stds-1788@xxxxxxxxxxxxxxxxx>
Sent: Tuesday, October 05, 2010 11:19 AM
Subject: Discussion paper: what are the level 2 datums?


P1788

I submit the attached paper for discussion, following which I aim to make it a motion.

John Pryce