Title: Two Views of ACCU's Conference

Francis wants to persuade more people to attend the annual Forum so he asked for people's opinions about this year's event in the hope that the voice of one of the 'erks' would be more persuasive than one of the association's brass hats. Here's my thoughts, feel free to use the green ink if you disagree.

First off. Francis wondered why only a sixth of the membership attended. Easy. Time and money. I was lucky, a bonus meant I could fund the two days out of my own pocket, I have family who live about an hours drive from Oxford so I had somewhere cheap to kip, I had annual leave to spare, my current projects aren't at the "only 48 hours till release and still full of 'features'" stage, a tolerably understanding partner and no children to deal with. Like I said, Easy. What happens if one or more of those conditions isn't met though? Let's pick two of them. Yes, a hundred quid for two days of this quality is cheap, but if you don't live within easy reach of the venue you have to find two nights accommodation, not just for yourself, but for your family as well then add travelling costs and suddenly it isn't a hundred pounds any more, it's little Johhny's fees for a year at Fettes/Eton/Cheltenham Ladies College (delete as appropriate). So, I wouldn't be surprised to find that a large number of members who wanted to attend hit the 'travel expenses' barrier and decided they couldn't afford it. Perhaps Edinburgh, Glasgow or Newcastle next year? Then there are the members working ridiculous hours to meet deadlines imposed by idiot managers (spot the tautology) and whose spouses are so inconsiderate as to want to have some time with the love of their life, even if it is just pushing the trolley round the supermarket together. Yes ACCU is a good thing, yes the Forum is excellent, but is it worth sacrificing a relationship or marriage for? Well, obviously I think it is, but I'm warped and I'm told I smell funny as well.

I hope that has saved a lot of aggrieved (private rather than corporate) members the effort of responding to Francis's (inevitable?) hectoring editorial on lack of interest (Where? Francis). Now to the conference itself.

Was it worth attending the C/C++ Forum? For me, yes. Did you miss out if you didn't attend? Well I guess it depends on whether you're an out and out guru or not. If you are then perhaps not, but if you're a guru why weren't you lecturing?

What did I get from it? I learned a number of new things, had a few ideas (prejudices) confirmed, was reminded that that there is a huge amount of C++ that I don't know. As if I need reminding.. I even managed to talk to a couple of people. Don't worry, doctors say they will recover. Given time.

What about the content? I opted for the C++ stream on Friday, with a short diversion into one of the C lectures so you'll have to rely on somebody else for the other streams. What follows is necessarily curtailed. Besides which, if you've got Web access you can get the presentation notes for yourself.

Like last year, I found this a thoroughly enjoyable event. The venue was historic if a little run down (the debating chamber must have seen better days) and some of the speakers were first class. The scope of the event had been widened since last year, and on the Friday tracks were available on Games Programming, Java and Component Technology, as well as on C, C++/OO and General Programming techniques. On the Saturday, the single track included two technical sessions and an hour discussing the standards process and progress, ending with a light-hearted language debate. On the Friday I attended the C++ track, so I can only comment on these sessions.

As a number of people unfamiliar with Oxford discovered, the how-to-find-it map provided was pretty hopeless, being topologically correct, but little more. On arriving at the Oxford Union, however, registration was efficient, the rather unusual selection of refreshments apparently limitless, and trial copies of several magazines (EXE, C++ Report, plus a few others) and software were there for the taking. My day of talks commenced with Julian Templeman's "Component Building with ATL," mainly because there was a gap in the C++ track. It was a fairly high-level introduction to AT, and was probably useful for those with no exposure to the ATL or Visual C++ 5 or above. The talk consisted mainly of screen shots of Microsoft's ATL wizard dialogs and a mention, but not detail, of the main ATL classes.

"Patterns & Techniques for C++ Resource Management," a two-hour session, found Kevlin Henney in fine form, as usual, in the Oxford Union debating chamber - a chilly and gloomy venue with uncomfortable seats (especially for the long sessions to follow). After a brief description of Patterns he mentioned the Command and Proxy patterns and the Smart Pointer Idiom in relation to resource management. Writing exception-safe code to avoid resource leaks was the next topic, and this highlighted the importance of getting the definition of operator=() correct for a class that manages resources. Reference counting was then introduced as a method of managing shared resources automatically, and a number of different strategies for this were outlined. Concurrency yields new dangers in programs, and synchronisation issues were discussed in terms of managing critical sections automatically as resources using the techniques discussed earlier. Finally, the general concept of "Resource Acquisition is Initialisation" (usually performed in the ctor) was discussed with a corresponding release of the resource in the dtor. For C++ developers, these issues are highly relevant. Techniques along these lines are almost unavoidable in the development of highly robust but complex software.

A relatively long lunch break allowed time to browse the exhibitors' stands. There were stands from companies such as Parkway Gordon (of course) to Inprise and IBM, QA training to QBS and Blackwells bookshop (which had very well stocked shelves). The conference raffle required delegates to visit each stand, and to prove this by obtaining a stamp on their raffle card. The wearied expressions on some exhibitors faces was evidence of the large number of trophy-hunters amongst the throng, eager to win from the impressive range of prizes on offer.

"The Standard C++ Library - Making it Work" by P.J. Plauger discussed the standard library from a library implementer's perspective. The "Making it Work" of the title implied getting it to function in the first place, rather than putting it through its paces. The library requires some of the latest compiler facilities, and so it was interesting to see how much work goes on behind the scenes getting the source to compile correctly on the available compilers. Next time you are cursing your compiler vendor for getting something wrong, be cheered by the thought that library writers have to tackle all these bugs in all the compilers! Dan Saks gave the next presentation, entitled "C++ Programming Style." In a characteristically epic three-hour session he covered a large amount of material concerning code writing style, documentation and design. Like last year's talk on the correct use of const, he gave an extremely thorough and insightful coverage of a topic that might initially seem to have little meat to offer. He started by describing exactly what he meant by saying some code was "better" than some other. The rest of the talk then applied this definition to a range of programming issues, beginning with code layout on the page, commenting (volume and location) and choice of variable names, through to scoping, encapsulation, accessors and mutators, function signatures and file organisation. (Anyone who came away without learning something must have dozed off in the dark. 8-)

For those only attending for the Saturday, the registration process fired up again. The rest of us trudged over to the chilly but now-familiar debating chamber. After the introduction was presented by a wooden chair in Francis' absence, "Idioms : Breaking the language Barrier," presented by Kevlin, covered a number of topics to do with OO, and related them to OO languages such as C++ and Java. Although initially pointing out how idioms that are fundamental in one language can be uncommon or irrelevant in another, the latter half of the talk described a number of idioms applicable to software development with examples in C++. Topics included copying semantics, iteration over collections, callback functions and value-based types, to name just a few.

After coffee, the "Standards Hour" explained where the C9X, C++ and Java standards efforts are at the moment, and where they are heading (in the case of Java, the response seemed to be "nowhere!"). A representative from each standards body presented the low-down on the standards process, and gave information on how to participate (the requirements for which - i.e. just turn up - surprised a lot of people). There were complaints about the cost of the standards documents once they are finalised - so prohibitive that they are lucky to sell a dozen copies in total! Apparently, history has been made, as an electronic version of a standard is now available for the first time - the C++ standard for $18 (visit http://webstore.ansi.org). This should mean that the standard can have a much wider audience.

After lunch, Dan Saks was on the podium again, this time discussing "The Impact of Namespaces in C++." And again, an apparently small topic was analysed in great and illuminating depth. Details and advice were given on basic namespace usage. More advanced topics were also covered, including nested namespaces, name collision resolution, unnamed namespaces and linkage issues.

Finally, the "Sacrificial Language Balloon Debate" had seven language proponents on the stage, each defending their chosen language from death by balloon eviction. With the audience a merciless 'clapometer,' languages such as Fortran and Beta were dispatched within minutes of the end of the initial presentations. With just C++ (Dan Saks), C (Neil Martin), Java (Russell Winder) and Forth (Francis Glassborow) remaining, Java and C took surprise early exits (Java first, then C), with C++ winning the day by a narrow margin. Forth's high ranking may, or may not, have been somewhat influenced by that language's representative.

For those that were not there, your editor was defending Forth and had to remind the audience that they might get a Forth Conference next year if they were not careful. And Dan Saks volunteered to jump from the balloon on the grounds that people's hearts were clearly ruling their heads.

So, in summary, a highly enjoyable two days, with a lot of good technical material upon which to ruminate, and some highly professional and well-prepared presentations. And at such a (relatively) affordable price, the bang per buck was sky high! Thanks to all those involved in organising it, and I for one hope I shall be able to attend again next year.