On Sun, Nov 16, 2014 at 01:20:45PM +0100, Maciej Sumiński wrote: > I am curious how clang would express its grief. They made a really > good progress on conciseness and clarity of error messages.
I don't think it's an issue in g++ itself... when template inference needs to go nesting down more than 25 levels *and* type names become consequently annotated the trail becomes longer... So I'd blame the trickery used by the boost people, not the compiler. Which is then due to language design... not that other 'modern' functional languages get it better as way to complicate a programmer's life (one word: monads :D) The worst thing is that obviously C++11 is *not* backward compatible with the 'current default' one (I guess C++99 or C++03?). That's bad for software maintenance: having to regularly port and 'fix' stuff made something like 20 years ago, hardware differences are more than enough; I don't need the language specification to change in the meantime:P Aside note: yes, we have 20yo software in production. Not of the PC kind, of course, but some people require spares and maintenance for 25/30/40 years after the release... (and pay it *a lot*) -- Lorenzo Marcantonio Logos Srl _______________________________________________ Mailing list: https://launchpad.net/~kicad-developers Post to : [email protected] Unsubscribe : https://launchpad.net/~kicad-developers More help : https://help.launchpad.net/ListHelp
