[GRASS-dev] Does g.parser accept 2-byte flags?

[Vaclav Petras]

On Sun, Nov 3, 2013 at 11:14 PM, Nikos Alexandris
<nik at nikosalexandris.net>wrote:

> Nikos Alexandris wrote:
>
> > > Is it possible to define a series of flags -cl -ch -mi -ma for a bash
> > > shell
> > > grass script, that g.parser will understand?
>
> Glynn Clements wrote:
>
> > No. G_parser() (and thus g.parser) follows Unix convention in its
> > interpretation of flags, i.e. "-abc" is equivalent to "-a -b -c".
> >
> > The double-dash flags (e.g. --quiet) are all hard-coded; it isn't
> > possible for a module to define its own.
> >
> > If single-character flags aren't suitable for whatever reason,
> > consider using a multi-valued option, e.g. "flags=cl,ch,mi,ma".
>
> The reason is ease in understanding (here preferred more than ease of
> use!):
>
> center low, center high, minimum modulator, maximum modulator, minimum
> modulator 2, maximum modulator 2. Only of low and high as well as only one
> of
> min and max can be defined at a time. And, then, only of min2 and max2 and
> only if a "-2" flag has been given.
>
> I currently use:
>  - l, h (which make sense for low, high
>  - i, a for m_i_n and m_a_x... yeah, brilliant idea... I know :-/
>  - n, x for mi_n_2 and ma_x_2... another brilliant idea :D
>
> You can define option value as ``minimum_modular_1`` but then write only
``minmod1``. The system is now same as for option keys. See latest
additions by Glynn described in *G7: parameter standardization: n/count,
min/minimum etc* [1].

I wonder how you dev's (strategically) select flag letters? Are there always
> some reserved letters?
>
> I personally use flags as a last possibility but of course they are
inevitable. My problem with flags is that if you see the module call, you
don't know what flag -s stands for (when I'm writing scripts with some
other programs I can use long versions, e.g. --standalone-version).

The other problem is that you don't usually know what happens if the flag
is not specified (e.g., what will be generated if "-s Generate standalone
version" is not specified). This is the same criticisms as exists for
checkboxes in GUI. This can be improved by good documentation/labels (e.g.,
"-s ...otherwise document stub is generated).

The last problem which I often see in GRASS modules and encounter myself is
that you by default want some behavior, let's say "Register maps in
temporal database". But here is the problem, how to disable this
functionality? Using -t flag is strange "-t Do not register maps in
temporal database" because -t as well as -r are suggesting action of
registration. We unfortunately cannot write --no-temporal or --n-t.

So, I don't use flags if possible and if I have to I don't see many chances
to do it right, except for writing extensive label and description
documentation. But maybe others see this more positively.

Vaclav

[1] http://lists.osgeo.org/pipermail/grass-dev/2013-October/065977.html
[2]
http://osgeo-org.1560.x6.nabble.com/G7-parameter-standardization-n-count-min-minimum-etc-td5082250.html
[3] http://comments.gmane.org/gmane.comp.gis.grass.devel/55256


> In any case, thanks for all of the feedback. Nikos
> _______________________________________________
> grass-dev mailing list
> grass-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-dev/attachments/20131104/647765d7/attachment.html>