LC-100 -- discouraging extensibility

Ref:  http://www.w3.org/QA/WG/lc-issues#x100

Submitter states:  "The position of the QA framework WG, that extensions 
should not be allowed, is quite clear. This is a political position, and 
doesn't accomodate those working on specifications that clearly demand 
public extensibility"

There are two errors in this statement:

1.) "[QAWG position is...] extensions should not be allowed".  This is 
inaccurate.  The specification [1] says: "Exercise caution in determining 
the extent to which extensions are allowed or not allowed. Since extensions 
can seriously compromise interoperability, specification writers should 
carefully consider whether extensions should be allowed."

It does NOT say that extensions should not be allowed, but it advises 
specification writers to carefully consider the often-negative impacts and 
implications, and make and document their decisions accordingly.

(The title of the issue "...discourage..." is more accurate than its 
statement -- we do discourage it unless it is well justified.)

2.) "This is a political position, and doesn't accomodate those working on 
specifications that clearly demand public extensibility"  This is also 
incorrect.  The entire extensibility content of SpecGL is based significant 
experience with standards and extensibility, including experience where 
unfettered extensibility has basically ruined the uptake of an otherwise 
good standard in the field.  It is NOT political, it is based on fact and 
experience.

Despite our bias against extensibility -- because it has often been used 
carelessly with disastrous interoperability impacts, or as an easy way to 
avoid hard decisions in writing the standard -- we nevertheless recognize 
that a total ban is inappropriate.  And that is reflected in the language.

IMO, Guideline 9 and its checkpoints strikes a decent balance, and I think 
what it clearly states is contrary to the premises of this comment.
There are some other specific extensibility issues which might lead to some 
fine tuning within the checkpoints.

Proposed resolution.  The comment misconstrues the actual specified intents 
of GL9 and its checkpoints, which intents are clearly spelled out in the 
2nd and 5th paragraphs.  Point this out to originator, no change to document.

(Note.  This does not mean that I think some of the language in the 
checkpoints cannot be fine tuned and improved -- I think that it can, per 
some of the other specific extensibility issues.)

Regards,
-Lofton.

[1] http://www.w3.org/TR/2003/WD-qaframe-spec-20030210/#Gd-extensions

Received on Friday, 4 April 2003 13:35:39 UTC