"Asger K. Alstrup Nielsen" <[EMAIL PROTECTED]> writes:
| On 4 Apr 2001, Jean-Marc Lasgouttes wrote:
|
| > Lars> lyxstring is a hack.
| >
| > Please don't be too impressed by all those big guys in commitees.
|
| I agree with Jean-Marc, Juergen, Allan, and the others.
|
| Lgb, this time, I think you are simply wrong.
about what? lyxstring is a hack? of course it is a hack! it is a hack
allowing us to use a close to conformant string class.
| An almost conformant, but really useful lyxstring is way better than a
| bloated, but maybe 99% conformant std::string. The few problems that
| lyxstring has in real life should be easy to fix.
mmm... make lyxstring play _well_ with stringstream.
how we do it today looks ok, but we go via a char* to do the
transformation stringstream -> string.
| Read the latest issue of C++ Users Journal for another take on the
| conformance argument. It seems obvious that the time is premature to
| let conformance be a main deciding argument these days. Issues such as
| compile time, footprint, and runtime performance are very much
| important, and will remain so for a long, long time.
Is it the conformance test by Herb Sutter you are refering to?
| If you don't care about those issues, you should not waste your time
| with C++, and go directly to Haskell or ML. Those languages are
| semantically well defined, contrary to C++, and most implementations
| are 99% conformant, compared to the lousy conformance of even the
| most conformant "state-of-the-art" C++ compiler. Also, it is questionable
| if there will ever be a 100% conformant C++ compiler, given the huge
| and at the same time vague semantics.
| Finally, C++ will never reach the same level of abstraction as those
| languages, so don't fool yourself.
bla bla etc.
| Abandoning lyxstring would be a huge regression at this point in time.
And who said we should do that?
--
Lgb
