Subject: Re: [boost] command line parameters parser
From: Vladimir Prus (ghost_at_[hidden])
Date: 2010-03-04 07:23:11
Denis Shevchenko wrote:
> Christoph Heindl wrote:
> > Hi Denis
> > How does your library compare to Boost.Program_options ?
> > -- Christoph
> >  http://www.boost.org/doc/libs/1_42_0/doc/html/program_options.html
> Hi, Christoph!
> I now about Boost.Program_options, and use it in my programs some time.
> But later I decide create my own solution, because Boost.Program_options
> have some limitation.
> First, it cannot call functions/functors. If I put in command line some
> parameter, I want some reaction on it. In my lib reaction occure
> immediately, in parsing process (but after all correction checks). Lib
> know nothing about this reaction's semantic, it just call correspond
Oh? Surely there's typed_value::notifier that allows to specify callbacks.
> Second, it cannot check "value semantic" of parameters. If parameter's
> value must be a valid path (or valid IPv4, or some else), I may want to
> check it immediately, in parsing process. In this case, my lib checks
> value's validity (using Boost.Filesystem, or my own algorithms) and, if
> error occurs, generate correspond std::exception.
It it unclear to me why you insist on doing such checks during the parsing.
ProgramOptions performns such checking when parsed tokens are stored
in variables_map, which seems like a sensible separation of responsibilities
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk