waynegitshell/gwsdat

GroundWater Spatiotemporal Data Analysis Tool (GWSDAT) R-Shiny implementation.

gwsdat's Issues

Load "Basic Example" data, if this is good, load "Extended Example".

Loading Comprehensive Example generates error in Contour plot.

• Not all wells are shown.

Display Spatial Contour Plot inside Shiny.

Needs adaptions to the panel variables, similar to the smooth time-series plot.

Think:

Which variables are shared and can go outside the plot function call, e.g. the 'DRV' variables?

test backlog card

How to simulate 'Subset Plot' Button for Traffic Light table?

Requires a subset selection of

• solutes
• wells
  Redraws traffic light table.

Add 'Traffic Light' Plot in 3rd tabPanel.

Get rid of TclTK panel (rp.control) for Shiny, but not Stand-Alone.

The function rp.control() creates the panel variable and a TK window. This is the basic call to create the TclTK window panel:

panel = rp.control(title = "", size = c(100, 100), panelname, background, ...)

Right now all the drawing related variables (DRV) are passed directly to this function (in the "..." part).
For Shiny I would not need rp.control. Instead only an object that holds all DRV, e.g.

DRV$Cont.Data
DRV$Well
etc.

To make it combatible with Stand-Alone I could pass this 'DRV' object to rp.control():

# Stand-Alone (TclTK)
panel = rp.control(title = "", size = c(100, 100), panelname, background, DRV)
# panel$DRV to access drawing stuff

and with Shiny I could just assign it directly to panel.

# Shiny
panel = 0
panel$DRV = DRV

Ommit plot() execution in GWSDAT_INIT().

These plots() are done while fitting. This is not necessary.

As long as GWSDAT_INIT() is not called inside a renderPlot() function, this isn't a problem, but still it is unnecessary overhead that might lead to problems on the server side.

Add 'Display Table' control into 'Smooth Time-Series' tabPanel.

This control allows to enable "Threshold" display.

It will be also added to the 'Traffic Light' tabPanel.

Pushing playback jumps 1 -> 2 -> 1 -> .. indefinitely after changing aggregation data.

Some data has not been set accordingly after the aggregation function has been called.

Display output from plot.GWSDAT.Data() inside Shiny UI, instead of TK Window.

Global variable 'pnl' is accessible from everywhere - all sessions can see it.

Put 'pnl' in session (user) context as soon as I've got users.

Make high-definition plots (see tk start option).

Display Smooth Time Series plot in shiny UI.

Add 'Data Aggregate' List box. Change Time Steps accordingly.

This requires to recall the aggregate data code located inside "GWSDAT Input Data.R".

Parse the code for the bits used for aggregation and make separate function that is called in the beginning (Init) and whenever the 'Data Aggregate' List box changes.

PlumeUnitHandlingFunc() not found when enabling 'PlumeDiagnostic' in Plot options.

Active tab: Contour Plot

Add controls to sidebar, same as in the tcltk panel.

Playback of Contour Plot faulty, jumps back and forth after slider is moved.

And the playback never stops.. maybe something with the time_step variable.

Show 'Color Key' image. See Button 'Color Key' in tcltk.

Show Pop-up image.

Show GW Contour lines do not show up when option actived in contour plot.

Add remaining components (traffic_light, groundwater flow, ..) to Init Script.

TRY OUT: Install R-Server on Linux and run Shiny on it.

See instruction here: https://www.rstudio.com/products/shiny/download-server

Note: the free version says it has no (1) user authentification and (2) no multi-processor support.

1. Is it still do-able with ptly and sqlite (see the User Authentificatin card) [How to link to another issue btw.?)

2. Serious issue considering that many users will access the service? Needs to be tested. Maybe there are other (free) server options.

Accumulate GWSDAT_Warning() objects in a list.

So several warning messages returned from a function can be displayed/evaluated.

Make simple httpuv script. Attempt to attach Shiny.

TRY OUT: shinydashboard for clean UI.

TRY OUT: User authentification with dply and SQLite inside Shiny.

Check out this: http://shiny.rstudio.com/gallery/authentication-and-database.html

No need for httpuv?

"Show GW contour" and GW flow arrows not displayed in contour plot.

Disappeared after adding aggregate function.

TRY OUT: using Drag-able Panels.

see here: http://shiny.rstudio.com/gallery/absolutely-positioned-panels.html

TRY OUT: Using tables that can be edited.

Applications:

• Edit input data.
• Edit parameters such as thresholds for each solute.

Execute TclTK windows stuff when GWSDAT is run stand-alone.

This will show the usual progress windows, main window, etc..

If GWSDAT runs with Shiny, we don't want these windows to be created. Instead, all tk*() status messages are passed back to the Shiny server.

Add Readme.md with basic information on running the App.

Add list with available data sets on Analyse page in Server Mode.

Pass error back from init script, instead of tcltk popup error message.

This involves all calls to tkmessageBox() inside GWSDAT_Run_shiny.R.

What is needed to run MakeSmoothPlot() on its own.

Right now it fails, the 'panel' object misses data (shadow.jjj, e.g.).

1. When the smooth panel in the main window is created it uses GWSDATpnl.
2. When this panel is ckicked it uses click.makeSmoothPanel(panel)

However, panel != GWSDATpnl.

Run contaminant fitting with push of a button.

Use simple push button

Plot_ImagePlot(): panel$tempTotalPlumeDetails saved but panel not passed back.

panel is updated inside plotting function, but currently Plot_ImagePlot() does not return the panel (unlike in GWSDAT MakePanel.R).

When I needed, update it. Better would be to recalculate panel$tempTotalPlumeDetails outside the plot function.

Consider using DT::dataTableOutput() - NOTE: this is just another table view , no editing.

This view is more compact (multiple tabs) and has different row shadings. Maybe better to look at.

Update smooth time-series plot as soon as any control changes value.

Insert 'Edit Thresholds' Button like control.

This affects Time-Series plot and Traffic Light table.

In tcltk it is done in a separate table, for each solute.

Reset button should clear everything in Import Data panel.

Import basic sample data into shiny script.

Do it manually to have the data available.

Insert data set information into "Analysis" Tab with option to jump back and select a data set.

Change of 'Time Step' slider should change time interval displayed in ImagePlot (contour) and Traffic Light table.

Make Sidebar more compact, i.e. 2 columns instead of just one.

GWSDATpnl object contain data, model, and draw (panel) related stuff. Need to untangle?

I need to separate data&model stuff from the drawing related stuff.

Right now rpanel calls use the rp.control() object GWSDATpnl that also includes data and model. But for shiny I would only need the model and data but not the drawing related stuff, since this is done differently in shiny. But confirm this, before doing any hard splits.

Threshold does not adapt to different concentration units.

As default threshold = 500 (mg/l). But when the unit is changed to ng/l or mg/l, the threshold value is not adapted but stays at '500'.

Insert data info and link to data on top of Analysis tab.

Click on Smooth Time-Series Plot opens image (png) window with Save button.

Insert "Import Data" Option, upload .csv via browser select window.

GWSDAT_Init(): Set values of 'dlines' and 'ScaleCols' (both checkbox ctrls) based on Default.Values if present.