Doc. no.   WG21/N2004=06-0074
Date:        2006-04-11
Project:     Programming Language C++
Reply to:   Beman Dawes <bdawes@acm.org>

Impact of Language Changes on LWG Schedule

The Library Working Group in Berlin reviewed all pending language change proposals from the standpoint of their potential impact on the LWG's ability to stick to the proposed schedule for the next standard. The relative merits of proposals were not considered - only the impact on schedule was discussed.

Two questions were asked of each potential language change:

• Would the proposed language change require or strongly suggest changes to the Standard Library interface or specifications?
   
• Even if not absolutely required, would the LWG like to utilize the proposed feature to improve the library interface or specifications?

If the answer to either of these questions was "yes", further questions were considered:

• How time-consuming do the potential library changes appear to be? A change might be considered time consuming for several reasons. It might impact large portions of the library, be technically difficult, or have a steep learning curve for LWG members.
   
• Must a potential library change be tested with a real compiler? Some interface changes are complex enough that the LWG won't be able to tell if they work unless tested with a real compiler implementation. Testing may also be required to ensure that interface changes do not unexpectedly break backward compatibility for user programs. (The LWG is willing to break backward compatibility after due consideration, but does not want to inadvertently break backward compatibility.)

If the answer to these questions was "yes", the proposed change has a possible impact on schedule.

Proposed language changes with possible impact on C++ 0x schedule

Very high impact changes

The following proposed changes will cause the C++0x schedule to slip if the committee does not commit to them by the end of the October, 2006 meeting, or if compilers implementing the language features do not become available to LWG members soon (say within the next year).

• R-value Reference.
• Concepts.
• Contract Programming.

High impact changes

The following proposed changes will cause the C++0x schedule to slip if the committee does not commit to them by the April, 2007 meeting. Although availability of test compilers is less critical, compilers supporting the bulk of the features are needed over the next 18 months.

• Auto and decltype.
• Lambda expressions and closures.
• Variatic templates.
• Template aliases (needed for TR2).
• Generalized constant expressions.
• New for-loop.
• Uniform Calling Syntax (Re-opening public interfaces).
• Extensible literals.

What does it mean for the committee to "commit" to a proposal?

The LWG doesn't have a single litmus test for the committee committing to a proposal. For a few, it might be just a report from the EWG that a proposal is likely to be brought forward. Quite a few members would like to see a proposal forwarded from EWG to CWG first. Some would like a report that CWG expects to have a formal motion ready soon. One or two LWG members want to see a proposal actually voted into the WP. Regardless of the details, LWG members do not want to spend serious effort on library changes, only to have a feature then not make it into the language.

Disclamer

This paper was prepared on behalf of the LWG. All errors and omissions, however, are my own.  --BGD