Re: [Liborigin-devel] Bug fixes

[Miquel G. <gbm...@gm...>]

2010/11/10 Knut Franke <Knu...@gm...>:
>
> What findStringPos() *should* do is guarantee that the read position does not
> change (i.e. seekg(startPos) before any return) and clearly indicate failure.
> There are different conventions for that (-1, a special value a la
> std::string::npos, returning bool or error code and writing the actual result
> to a pointer as offered by the GSL, throwing an exception). I think it would
> help if we could try to stick to one such convention.

Yes, you are right, but, since the generalized parser will not  need
findStringPos()
at all, I saw no need to enhance it. You should not return -1 in an
unsigned int,
unless you use it to represent the largest available unsigned int. On
the other hand
 <cstring> header uses  string::npos = static_cast<size_type>(-1).
In the case of findStringPos, I agree that a better patch would be:

size_t Origin800Parser::findStringPos(const string& name)
{
...
return std::string::npos;
}
since to represent a position is what size_t is designed for.

However, I would better spend time in the generalized parser that
finding a good  patch
for each bug in OriginNNNParser.

Miquel