C/C++ Users Journal June, 2004
First, a confession. I'm writing this editorial in the aft cabin of a motor yacht cruising the Whitsunday Islands along Australia's Great Barrier Reef. (This is the "gin palace" that I alluded to last month, and it is indeed well equipped in the gin department.) Tana, Geoffrey, and I are the pampered guests of John O'Brien, head of Whitesmiths Australia, and his wife Kate. We're enjoying a luxury that I don't expect to become accustomed to. More important, Tana and I are unwinding after several weeks of Standards meetings, in the company of our son and some old and dear friends.
We began in Melbourne with a three-day meeting of ECMA TC39/TG5, hosted by Monash University. TG5 is the bunch chartered with developing a Standard for Managed C++, Microsoft's extension to Standard C++, to keep it a player in the era of C# and .NET. This meeting profited from the presence of a couple of WG21 regulars, to widen the bandwidth with the Standard C++ folks. I can report that we made good progress in ironing out some remaining design issues, but we have much to do to keep to the ambitious schedule of getting an approved Standard by the end of this year.
Then it was up to Sydney for a week of WG21 (C++). Strictly speaking, we were in Coogee (KUH-jee), a funky little beach town a few kilometers southeast of Sydney proper. Whitesmiths hosted the meetings at a hotel just across from the beach. The Library Subcommittee is still on target to produce a nonnormative Technical Report that will add all sorts of extra libraries to Standard C++. That includes the many functions added to Standard C with the 1999 revision (C99). And the whole committee voted strongly in favor of reviewing all the language features added with C99, with an eye to minimizing dialect differences. Finally, I'm happy to report that Dinkumware's Pete Becker assumed the role of Project Editor, with the principal responsibility of editing and maintaining the draft C++ Standard as it is revised.
WG21 was followed by a week of WG14 (C), also in Coogee. We had brief, friendly discussions about how to continue to develop the special math functions and decimal floating point in conjunction with WG21. We put in serious time on the Secure C Technical Report advanced by Microsoft. We managed to review *all* of the open Defect Reports. And we even gathered the most stable Technical Corrigenda into a TC2 so we can issue a tidied up C Standard by 2005. (Don't worry, these are all niggling changes and clarifications. If you bought the book on C 2003, with TC1 applied, published recently by BSI, it'll still be an excellent reference.)
Don't get me wrong, the meetings were not all smiles and sunshine. Both committees had some difficult discussions that grew heated at times. But the overall spirit of good will and cooperation between C and C++ that I reported after the Kona meetings still prevailed.
Perhaps the biggest concern aired at both meetings is the growing difficulty in finding hosts for future WG14/WG21 meetings. Having the two committees meet back-to-back obviously doubles the time investment. For hosts primarily interested in C, it rather more than doubles the total cost as well. We can't always find a host as accommodating as Whitesmiths Australia, which even paid for lunches as well as a dinner reception for each group. John O'Brien may choose to be generous with his friends, but he gets little corporate benefit out of hosting either C or C++ Standards meetings, let alone both.
A further irony is that many recent meetings have been hosted by the smallest of companies. Plum Hall has been notoriously willing over the years to pick up the tab for meetings on the Big Island of Hawaii, a very attractive venue to say the least. Perennial, Edison Design Group, and, yes, Dinkumware have all stepped forward on several occasions of late. Let me hasten to add that Microsoft has been quite generous in this department. But there are quite a few companies, which I won't discomfit by naming, in that vast middle ground who arguably have a stake in the advancement of these programming language Standards.
I've never hesitated to grumble when I think that either committee has gone in a wrong direction, but I have to observe that the C and C++ Standards have been two of the most successful in history. They both serve to reduce dialect splintering (at least, in the medium to long haul). And the ongoing work of both committees has provided a forum for the controlled evolution of two very important programming languages. It would be a pity to see this effort run down and die for want of proper support.
P.J. Plauger
Senior Contributing Editor
pjp@plauger.com