In this article, Scott Meyers shares his picks for the five most important non-book publications in the history of C++, along with why he chose them.

http://www.artima.com/cppsource/top_cpp_publications.html

What do you think of Scott's choices? What other C++ publications do you feel have been important in the history of C++, and why?

I miss Guro of the Week from the list. Though it may count as a different kind of "publication" that didn't play in this category.

To be honest, I didn't consider the GOTW series, but had I, it might well have made the list. I can't think of any single GOTW article that alone would be important enough to dislodge any of the entries I listed, but the series as a whole might have, especially considering the impact of the books that were spun off from the series.

For me C++ articles about resource management are the most important, eg.

Bartosz Milewski: Resource Management in C++. JOOP 10(1): 14-22 (1997)
Tom Cargill: Managing Dynamic Objects in C++. DDJ Jul 22 (2001) http://www.ddj.com/184409895

I think we may systematically underestimate the importance of "non-publications" such as websites, newsgroups, and other web sources. consider comp.lang.c++ and alt.comp.lang.learn.c-c++. Does anyone have hard numbers and ways of calibrating?

For years (almost 7 years now), "Learning Standard C++ as a new language" averaged 100 downloads a day. That's not my most popular "page" either. This adds up. I suspect that less conventional, less "serious", and more responsive sites may have overwhelmed the conventional sources of information.

Marshall Cline's C++ FAQ Lite on the web is an extremely useful resource which contains best practices and very helpful advice.

http://www.parashift.com/c++-faq-lite/

That and comp.lang.c++.moderated are my two non-book resources that I use the most.

As for exception-safety language in the standard...

The nothrow guarantee is expressed as "Throws: Nothing ..." and the strong guarantee is expressed as "If an exception is thrown ... there are no effects.", where the ellipsis contains any "unless" or "except" clauses needed. This wording was chosen to fit with how the standard library was already described, and to minimize the number of edits to a document that was supposed to be almost done.

The basic guarantee for things in the standard basically amounts to "X does what the standard says it does." It applies to everything in the standard, for lack of any permission to violate it, and so goes unsaid.

As terse as it is, the standard is still too long, so introducing non-normative definitions of terms not actually used was a nonstarter.

And so it came to be that the phrases "strong guarantee," "basic guarantee," and "nothrow guarantee" don't appear in the standard. Not all that interesting, but there you have it.