Re Bjarne Stroustrup's list, did anybody notice that the most important C++ person's name is missing from the list?

Now that is the humility of the truly great that I was referring to earlier.

> In second place behind Stroustrup I'd put The Humble
> Programmer. He has kept C++ alive for so long despite its
> numerous deficits.
>
> BTW, after so much self-adulation why not continue the
> series with "The Most Important C++ Failures...Ever".
> Candidates are (among others):
> - Standard compliance of compilers
> - binary compatibility
> - module support (lack of)
> - 'C++ will absorb C'
> - multi-paradigm induced complexity
> - templates
> - MFC
> - STL
> - BOOST
> - ...

Are you just trolling, or would you care to come with some justification of why you call the above "failures"?

I'm particularly interested in the part about the multi-paradigm design of C++, templates, STL, and Boost, which I think are all important contributions of C++ (or to C++), and to the field of computing.

I found the following people missing from the list.

Krzysztof Czarnecki and Ulrich Eisenecker

> I found the following people missing from the list.
>
> Krzysztof Czarnecki and Ulrich Eisenecker

Thanks

One could add some magazine columnists (Pete Becker, ...) some heavy and high-quality newsgroup posters (James Kanze, ...) and some C++ book authors. But no matter how many people you add there will always be someone missing.

> For the longest time I began every phone screen for C++
> talent with "Who is Scott Meyers?". Failure to correctly
> answer would get you 10 more minutes of pleasant talk and
> a "thank you for your time".

Well, that sounds like a good strategy for eliminating applicants, but not a very good one for selecting competent C++ programmers. Certainly with the hundreds of books and other resources available to learn and master C++, being unfamiliar with the work of any one author shouldn't be a fatal flaw.

> I still routinely see violations of Scott's 50 items in
> recently written production code and it gives me the
> willies everytime.

I'm not a big fan of these rule lists. Individuals usually have a rather narrow range of experience to draw from and so they may not realize there are environments unfamiliar to them where the rules are "wrong".

> Are you just trolling, or would you care to come with some
> justification of why you call the above "failures"?

The list contains 'heavyweights' that had some impact on C++ but also failed in critical aspects (while succeeding in others). I'd expect one of the most important C++ people ever to also discuss those failed aspects.

I think the the most important reason for success of C++ is the brain power behind it - the number of smart, knowledgable and entusiastic people within C++ community is truly amazing.

Therefore, maybe this article should be about the 50 most important C++ people ever :)

The most important woman missing from all of these lists is Josée Lajoie.

> > For the longest time I began every phone screen for C++
> > talent with "Who is Scott Meyers?".
>
> Well, that sounds like a good strategy for eliminating
> applicants, but not a very good one for selecting
> competent C++ programmers.

With 100 resumes for a given job - that's the idea.

> > I still routinely see violations of Scott's 50 items in
> > recently written production code and it gives me the
> > willies everytime.
>
> I'm not a big fan of these rule lists.

Well I am. It is astonishing how many so-called C++ programmers see no problem subclassing a class with a non-trivial non-virtual dtor. That's just one example.

The fact is, that the vast majority of people who code in C++ aren't really qualified to do it. And I'm tired of working with those people - so I don't do it anymore.

> With 100 resumes for a given job - that's the idea.

Of course you can use any method you choose. My point is that you may be eliminating some very competent people. If you think the book is important why not ask candidates questions based on the content rather than assuming they can't know the answers unless they've read the book?

> > Are you just trolling, or would you care to come with
> some
> > justification of why you call the above "failures"?

> > I'm particularly interested in the part about the
> > multi-paradigm design of C++, templates, STL, and
> > Boost, which I think are all important contributions of
> > C++ (or to C++), and to the field of computing.

> The list contains 'heavyweights' that had some impact on
> C++ but also failed in critical aspects (while succeeding
> in others). I'd expect one of the most important C++
> people ever to also discuss those failed aspects.

A feature having problematic aspects doesn't in itself make the feature as a whole failed. What I reacted to was that much of the list is considered successes in the community.

For a more productive discussion, maybe you could list some of those "failed aspects", then we can discuss them here?

> Well I am. It is astonishing how many so-called C++
> programmers see no problem subclassing a class with a
> non-trivial non-virtual dtor.

Would it matter if that non-trivial non-virtual constructor was protected...?

> The fact is, that the vast majority of people who code in
> C++ aren't really qualified to do it. And I'm tired of
> working with those people - so I don't do it anymore.

I certainly understand that sentiment. The the thing I'm wondering, then, is: Are you able to find as good people as you want, for positions?

> It is astonishing how many so-called C++
> programmers see no problem subclassing a class with a
> non-trivial non-virtual dtor. That's just one example.

You mean like
struct input_iterator_tag {};
struct forward_iterator_tag: public input_iterator_tag {};

> The fact is, that the vast majority of people who code in
> C++ aren't really qualified to do it.

Blame C++, not the people.

> And I'm tired of
> working with those people - so I don't do it anymore.

So you are not working with C++ anymore?

> Blame C++, not the people.

Oh I do, but we have this legacy code that needs maintaining, see. Frankly, the project would have been better done in Erlang.

> So you are not working with C++ anymore?

It is a language for a madman and overly labor intensive besides. I now personally only code in dynamic languages. All current web development for personal work is done in Squeak/Seaside (web) or Objective C (desktop app). If I need speed, I use bits of C.

To quote Tom Cargill: "what is a protected abstract virtual base pure virtual private destructor and when was the last time you needed one?" pretty well sums it up. I think he has abandoned the language as untenable as well (focused on Java now).

As for work - I manage developers - I no longer code in it. I interview hundreds of candidate developers for several languages every year - in my experience most of the C++ candidates are dangerous. My personal skill level was Grade A for C++ circa 1997. It is a little out of date now I think - but I don't see any compelling reason to use it outside of maybe device drivers (and then I'd just use C).