Re: gsl_histogram*_clone should fail silently

[Brian Gough <bjg at network-theory dot co dot uk>]

Achim Gaedke writes:
 > Sorry for that ambiguous subject. Of course GSL_ERROR should be
 > used. Before suggesting that patch I assured that the error handler
 > is called in every case of failure.

Originally it was dicussed whether to have only one call to gsl_error
per function call, in the way you suggested, or to allow multiple
calls. I decided that it was simpler to do things the way they are now
--- any function can make multiple calls to gsl_error before returning
(even if it does not call other gsl functions). So exception handling
code needs to take that into account -- the easiest way is to log the
first error and ignore subsequent errors until a flag is cleared.

Brian