From: boat042-nospam@yahoo.com   
      
   James K. Lowden wrote:   
   > On Fri, 10 Feb 2012 14:29:50 -0800 (PST)   
   > Richard Smith  wrote:   
   >   
   >> What I'd really like is something, preferably on-line, that I can   
   >> point the other developers at that demonstrates that (i) the standard   
   >> library does work.  Really.  And (ii) it's fast enough.  Or, for that   
   >> matter, any other ideas on how to get the other developers on side.   
   >   
   > Herb Sutter wrote some articles on the STL making the case that you   
   > simply can't outperform the STL in general.  In some specific case you   
   > might see some slight advantage, but not in general.   
   >   
   > I wouldn't make the efficiency case, though.  I'd make the case for   
   > laziness and correctness:   
   >   
   > 1.  Code you don't write is code you don't maintain.   
   >   
   > 2.  The probability of bugs in the STL is orders of magnitude less   
   > than in your own code.   
   >   
   > 3.  Using the standard library whenever possible improves the code as   
   > a means of communicating to other *programmers*.  It speeds   
   > comprehension and promotes standard idioms, which are also less   
   > likely to contain errors.   
   >   
   > 4.  It lowers the barrier to entry.  If just one person is attracted   
   > to the project because the code is readable, the benefit to the   
   > project will outweigh any putative benefit of non-standard constructs.   
   >   
   > Keep in mind, though, that people get involved in programming projects   
   > to write programs.  Sometimes that's an end to itself, secondary to   
   > the project's stated goal.  Sometimes pride of authorship -- or   
   > the desire to work at a lower level than one would otherwise be   
   > allowed to -- can lead to the kinds of arguments you describe.   
   >   
   > For many years "programmer productivity" was measured in lines of   
   > code.  Djisktra used to ridicule that, pointing out that code   
   > was a cost, not a benefit.  Management was not just using a poor   
   > metric, but "had the number on the wrong side of the ledger"!   
      
   All are good points, of course. The problem really isn't one of politics so   
   much as it is of logistics. How long does it take to examine, update,   
   convert, and regression test 250k lines of code? It's a lost cause already   
   if you even entertain notions that it's about the politics of the old-guard   
   versus new. With broad acceptance of C++11 looming, many of us will be   
   facing a similar question very soon.   
      
      
   --   
         [ See http://www.gotw.ca/resources/clcm.htm for info about ]   
         [ comp.lang.c++.moderated.    First time posters: Do this! ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)