[GRASS-dev] No Required GUI section for g.list and g.remove

[Vaclav Petras]

On Tue, Oct 28, 2014 at 10:31 AM, Moritz Lennert <
mlennert at club.worldonline.be> wrote:

> On 25/10/14 02:58, Anna Petrášová wrote:
>
>> Hi,
>>
>> On Tue, Oct 14, 2014 at 11:56 AM, Anna Petrášová <kratochanna at gmail.com
>> <mailto:kratochanna at gmail.com>> wrote:
>>
>>     Hi,
>>
>>     On Tue, Oct 14, 2014 at 8:58 AM, Vaclav Petras <wenzeslaus at gmail.com
>>     <mailto:wenzeslaus at gmail.com>> wrote:
>>
>>
>>
>>         On Tue, Oct 14, 2014 at 5:28 AM, Moritz Lennert
>>         <mlennert at club.worldonline.be
>>         <mailto:mlennert at club.worldonline.be>> wrote:
>>
>>             On 14/10/14 10:38, Paulo van Breugel wrote:
>>
>>
>>
>>                 On Tue, Oct 14, 2014 at 10:06 AM, Moritz Lennert
>>                 <mlennert at club.worldonline.be
>>                 <mailto:mlennert at club.worldonline.be>
>>                 <mailto:mlennert at club.__worldonline.be
>>
>>                 <mailto:mlennert at club.worldonline.be>>> wrote:
>>
>>                      On 14/10/14 09:14, Paulo van Breugel wrote:
>>
>>                          Putting the 'ignore' option in  separate tab
>>                 with patterns is fine I
>>                          think. Also, for g.remove to have the 'type'
>>                 and 'name' together
>>                          in one
>>                          tab is also a good idea imho.
>>
>>                          I am not sure I understand the last question;
>>                 you mean to add the
>>                          possibility to make an option required but
>>                 still have the option
>>                          to put
>>                          it in another section? I think that would be a
>>                 good idea, not
>>                          only in
>>                          this case, but more in general, it would make
>>                 it easier to create an
>>                          consistent interface for modules that require
>>                 more than a few
>>                          inputs.
>>                          It might be a good idea to flag options as
>>                 required, e.g., by adding
>>                          '(required)' after the option name?
>>
>>
>>                      I'm not sure I agree with this as this would leave
>>                 the door open for
>>                      required flags to be disseminated across several
>>                 sections. I like
>>                      the fact that the use immediately sees what is
>>                 required to run the
>>                      module.
>>
>>
>>                 I guess it is a matter of preference. There are some
>>                 grey area or cases
>>                 where this separate 'required' tab does not really work
>>                 i.m.h.o.
>>
>>                 The '-f' flag in g.remove is one example. You can run
>>                 the module without
>>                 (so it shouldn't go in the 'required' tab), but you
>>                 can't do what the
>>                 module is basically meant to do without it, which is
>>                 removing layers (so
>>                 in that sense it should be a required choice).
>>
>>                 Perhaps a better example is r.random. One of the
>>                 required options is the
>>                 output layer. That can be a raster layer, a vector layer
>>                 or both.
>>                 Because of this construct, the required output name
>>                 cannot be marked as
>>                 required. Solution is to use a separate tab 'optional'
>>                 where the user
>>                 can provide the output name of the vector, raster or
>>                 both layers. So the
>>                 user has to fill in required information in a 'required
>>                 tab' and an
>>                 'optional' tab. I don't think it is really problematic
>>                 as failing to
>>                 give the output name results in a clear error message,
>>                 but it isn't
>>                 exactly consistent.
>>
>>
>>             The new options to declare parameters as mutually exclusive
>>             or as "at least one is requried of the group" might be a
>>             solution to that, but no idea how to implement this in the
>> GUI.
>>
>>         Put this to GUI is certainly needed but challenging and I it
>>         will not be included in 7.0. Perhaps we should put this to the
>>         manual in some way. But modules are not using it anyway.
>>
>>         Anyway, these "at least one required" are causing Required
>>         section to be less and less used, so that's another reason why
>>         it makes sense to leave it out sometimes.
>>
>>
>>     I think it makes thanks to 'override' the required property with the
>>     guisection. If we do it for a module, we should make sure there is
>>     no Required tab at all. I think having required parameters in custom
>>     tabs and eliminate Required tab is totally fine. Having Required tab
>>     and at the same time have required parameters in other sections
>>     would not work well.
>>
>>
>>     Also we could mark the required options in the gui somehow, for
>>     example add a red star? In the code I see attempts to render the
>>     labels as bold, it is not used eventually, but I don't think bold is
>>     the best way anyway.
>>
>>
>> I attached screenshots with using the red star for required options (and
>> in this case I was also testing when the guisection is preferred to
>> required). In my opinion, it gives you good idea what is required or
>> not. There are some problems coming from the wxPython limitations, for
>> example the one which you see on the second screenshot: when the label
>> is part of the border, I can't set the asterisk red, the color can be
>> changed only for the entire label. But for majority options, it works.
>>
>> Regarding the required vs. guisection, I really think we should try to
>> organize the options logically, not based on required/optional. Some
>> distinction of the required options is then needed and the red asterisk
>> seems to be a good solution. But we can discuss some other options too,
>> of course.
>>
>
> If you really want to go down that path (it would get a 0 from me), then
> please remove the required tab completely. It will be very confusing to
> have a required section, but not with all required parameters in it.
>
> Right. The modules which have required options in some other sections than
Required, should not have the Required section.


> I personally like the way the required tab focuses attention on the most
> important parameters, even though I do understand that there are
> ambiguities that this system creates. I'm not sure, though, that getting
> rid of the required tab is the solution. But if you go in that direction it
> should be all the way.
>
> We are not getting rid of it completely. If it is appropriate for the
module, it can use it. It is still the default. The point is that for some
modules there is no Required tab because of different ways to specify
inputs (e.g. map names in file vs map names in parameter) and for some of
the modules there is more logical way of organizing parameters than
Required vs optional (Optional section or general sections). Enabling
required parameters outside Required section is addressing the later case.

Moritz
>
> Moritz
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-dev/attachments/20141028/30cf1b19/attachment.html>