[libcamera-devel] [PATCH] cam: options: optional arguments needs to be specified as --foo=bar

[Niklas Söderlund]

Hello,

On 2019-01-25 16:44:41 +0200, Laurent Pinchart wrote:
> Hi Kieran,
> 
> On Fri, Jan 25, 2019 at 12:23:25PM +0100, Niklas Söderlund wrote:
> > On 2019-01-25 13:01:34 +0200, Laurent Pinchart wrote:
> > > On Fri, Jan 25, 2019 at 12:23:11AM +0100, Niklas Söderlund wrote:
> > > > It's not state in the documentation but optional arguments needs to be
> > > > specified using as '--foo=bar' instead of '--foo bar', otherwise the
> > > > value is not propagated to optarg during argument parsing. Update the
> > > > usage printing helper to reflect this requirement.
> > > > 
> > > > Signed-off-by: Niklas Söderlund <niklas.soderlund at ragnatech.se>
> > > > ---
> > > >  src/cam/options.cpp | 3 ++-
> > > >  1 file changed, 2 insertions(+), 1 deletion(-)
> > > > 
> > > > diff --git a/src/cam/options.cpp b/src/cam/options.cpp
> > > > index 82acff9bbeea476d..73d81d0bc0ec6d38 100644
> > > > --- a/src/cam/options.cpp
> > > > +++ b/src/cam/options.cpp
> > > > @@ -143,7 +143,8 @@ void OptionsParser::usage()
> > > >  		};
> > > >  
> > > >  		if (option.argument != ArgumentNone) {
> > > > -			argument += std::string(" ");
> > > > +			argument += option.argument == ArgumentOptional ?
> > > > +				"=" : " ";
> > > >  			if (option.argument == ArgumentOptional)
> > > >  				argument += "[";
> > > >  			argument += option.argumentName;
> > > 
> > > This will output
> > > 
> > > 	-f, --foo value
> > > 
> > > for mandatory arguments, and
> > > 
> > > 	-f, --foo=[value]
> > > 
> > > for optional arguments. If we want to print the =, shouldn't it be
> > > --foo[=value] ?
> > 
> > It should of course be --foo[=value], will fix.
> > 
> > > And how should we handle the case where no long option
> > > is available, with this patch -f=[value] would be printed, which isn't
> > > correct I think.
> > 
> > Good point, for optional short arguments the syntax would be -fvalue.  
> > Would it make sens to print both short and long syntax in the usage?
> > 
> >     -f[value], --foo[=value]
> > 
> > Let me know what you think and I send a v2.
> 
> That would make sense, but then we should also write
> 
>     -f value, --foo value
> 
> and we will lose the nice alignment of all long arguments :-( I'm not
> sure what's best.

To move forward with this I would suggest,

    -f, --foo[=value]

Rational being that if you use -f you get the default behavior and if 
you want to specify the optional argument you can use the long option. I 
would however not go out of the way to make -fvalue fail argument 
parsing. Then when we create a man page for the tool we can specify both 
versions in detail.

Would this work for everyone?

> 
> -- 
> Regards,
> 
> Laurent Pinchart

-- 
Regards,
Niklas Söderlund