JTC1/SC22/WG21
N0814
Document: WG21/N0814=x3j16/95-0214
Date: 09 November 1995
Author: Nathan Myers
myersn@roguewave.com
Why We Should Not Break the Locale Invariant ("Yet")
----------------------------------------------------
+ We had agreed not to pass controversial or fundamental changes
in Tokyo, in the absence of many of the members who regularly
attend meetings. I believe Bill's is the only such proposal
offered at this meeting.
+ The vote in the working group cannot be considered a vote in favor
of the proposal, because it had not (and indeed has not) been written.
The group voted in favor of Bill's approach, based on oral arguments
which were not published and to which there was no opportunity to
prepare a response. We voted in favor of seeing a complete, detailed
proposal. What he has produced, thus far, is not that, but only
a list of working paper edits. This is not enough to support a
informed vote on the matter.
+ The normal rule of the committee is that the advocate of a
substantial change has the burden of identifying the consequences
and demonstrating how any problems that arise can be mitigated.
All of us who have had substantial motions passed have done so.
Bill has not done that homework.
+ I plan to write a detailed analysis of the consequences of
this proposal; I have been unable to drop other work this week
to concentrate on such writing, nor opportunity to prepare
such an analysis beforehand.
+ The primary argument for this motion has been an argument that
breaking encapsulation makes some programs easier for some
people; as always, encapsulation always makes some things less
convenient, but with consequent architectural benefits. Locale
issues are *not* fundamentally different from other architectural
matters. Nothing in the status quo prevents using global locales;
but we are not obliged to cater to such use at the expense of those
who benefit from encapsulation.
It is an easy courtesy to extend, both to the other committee members
who (due to financial constraints) could not attend the Tokyo meeting,
and to me as the author of the locale components, to defer a vote on
this proposal and allow all of us time to study and respond to it,
and perhaps arrange a compromise. Many other issues have been deferred
for much less compelling reasons.