eWeek discusses C++ futures

Esther Schindler of eWeek wrote on this week's C++ Connections conference.  Near the end of the article, she noted the hullabaloo surrounding our use of the word "deprecated" to warn on the use of potentially unsafe standard library functions.  Yes, the word "deprecated" in this context is incorrect, and -- as Herb mentioned -- we intend to change it.  In the interest of complete transparency here, this subject did come up prior to VS 2005 release, and we considered changing the word to something like "unsafe," but it came up at a stage in the cycle that we were uncomfortable with the potential destabilizing effects of changes necessary to properly address the issue.  Yes, the wording is wrong and perhaps even rude, but the fact is that the current behavior doesn't change the meaning or behavior of your code and the warning can be disabled, so it wasn't deemed worth the risk at that point in time.  It's just one of the hundreds of tough calls that need to be made as a software project nears the wire.

Comments

• Anonymous
  November 10, 2005
  Yeah, "deprecated" is incorrect because the standards committee knuckled under to unseen forces, but "should have been deprecated 30 years ago" would be correct. 20 years ago some wise engineers tried to persuade the committee to deprecate them in the first version of the standard, still being included by necessity but with hopes of excluding them later. I don't remember what year was the CACM article on string handling in C but it was surely more than 20 years ago.
  
  Even though C in those days was still a replacement for assembly language, it would have been better to choose a different machine's assembly language as a model for C's string library. There were some machines that stored counted strings. There were some where the string count was known by the assembler (even if not stored together with the string at runtime) and executable instructions included the count.