Hello all, A quick summary view has been implemented which shows data section title, abstract, geographic extent, spatial representation type and reference system. For the name of reference system I'm using ReferenceSystem.getName().toString() but it may not always return a proper name like what you can see in the tree table view reference system info section. That is one problem so far.
Secondly axis dimensions have not yet been added. In one file I have, there are two spatial representation info sections with two axis dimension properties each. Would you prefer to see them grouped in the separately or all together in the table, though the order is guaranteed to be maintained? I have added an fxml screenshot for the CRSEditor. I have also added an area description and bounding box view. Do comment on the format and any changes. Regards Siddhesh Rane ---- Martin Desruisseaux wrote ---- >Hello Siddhesh > >Le 28/07/2017 à 09:22, Siddhesh Rane a écrit : > >Are you pointing out to the non uniform widths of various spinners and combo >boxes? These can be fixed by setting their maximum width to max value so >they'll fill the entire width of the column. > >I was thinking more about the diversity of layouts. One row may have 2 fields >for a minimum and maximum values, another row 4 fields, etc. In tables, data >in the same column are usually represented in an uniform way. This is not >completely true for metadata so we are taking some liberties with specialized >widgets, but I think that we should not go too far. I suggest to use in the >tree-table only widgets representable on a single line, and move larger >widgets outside the tree-table (more on it later). > > >Actually we do. I have put labels in the latest update. We can have as much >vertical space as we want. Earlier I was just using prompt text but now we >have both. > >Yes, those labels help a lot. Thanks. > > >I'm not fully clear about what constitutes a simplified view. My view was that >a simplified view will group together certain nodes that are related like min >max, bounds, date and date type etc and show hide empty nodes and certain user >specified nodes. But I thought the hierarchy had to be maintained because it >has useful information, as certain fields like Identifier and citation appear >at many different places, so I stuck to tree view. > >My idea of simplified view is to get ride of the hierarchy. Indeed it is not >possible to represent the full ISO 19115 metadata without hierarchy. As you >point out, structures like Identifier and Citation appear in many different >places. But the proposal is to pick only one Citation and only one Identifier, >the ones which are generally of most interest to users. For example an ISO >19115 metadata can contain Citation for: > >datametadata (may or may not be same as data citation) >file formatalgorithm of data processing >process for quality assessmentetc. > >But most of the time the user only wants to see the citation for the data, so >we could show only that citation in the simplified view. If (s)he want to see >other citations, then (s)he can switch to the tree-table view where all >information are available. > >One of the first things that users will want to see is the title of the data >and the geographic extent. In the tree-table view, users have to navigate into >"Identification info" node, then "Citation", then "title" (and yet deeper if >(s)he wants the author name). For the the geographic extent, users have to >navigate to "Identification info", "Extents", "Geographic elements". This may >not be obvious to users not familiar with ISO 19115. The proposal is to have a >summary panel without hierarchy, showing only a pre-selected metadata (not >all). > > >According to the the sample image that you sent me I have come up with a >similar summary view in tree table view instead using operations like removing >unwanted subtree, flattening subtrees and rearranging nodes at >https://github.com/SiddheshRane/sis-client/blob/master/screenshots/Summary.png >This view was generated by user interface elements like shift+up/down for >rearranging nodes, and right clicking to flatten subtree or remove nodes from >the view. Do you want the simplified view to show only titles, names, >descriptions, keywords, strings? Like You said Citation could be hidden but >show only it's children, what other components are less important. > >I would suggest that the left column disappear for the simplified view; no >tree at all. So the user would see "Title" directly, without "Identification >info" node on top of it (but a button somewhere would hello him to switch to >the tree-table view if (s)he wants). I would omit the online resources and >contact information for now (we can revisit them later). For a first draft, I >would suggest to use only the following: > >Title on one line >Abstract on many linesGeographic extent >Spatial reference system (only the name, not the full description) Spatial >representation typeAxis dimensions in a table > >Nothing more for now (we can revisit later for more info). Above can be >read-only (I think it is okay to ask users to switch to tree-table view if >they want to edit). > > >Oh I didn't know about the linear relationship. I will switch to Plate-Carré >projection image. > >Alternatively, we could also omit the range sliders. They are not linearly >related to the selection unless we restrict to Plate-Carrée, which is not the >preferred projection. Even in Plate-carrée those sliders are not aligned with >the selection unless the user zoom at a very specific level. All this come at >the cost of an extra dependency. Wouldn't it be preferable to remove them for >now? If we want range sliders for geographic extent, we will need to develop >our own, but this would be more work than what we can do in the GSoC. > > > >Can you tell me the requirements of the CRS editor, like how do you wish to >interact with it and what are the operations you need to perform. Whether you >only need to edit certain values, or add new components like axes, parameters >or change the types entirely like switching from geodeticCRS to projectedCRS. >I need your requirements clearly, even if certain questions might sound >senseless, because I'm not an expert at these. > >I suggest to begin with a dialog box like below (for easier start, I'm >omitting some important elements like base CRS, ellipsoid name, etc.): > >The CRS name as a text fieldThe datum name as a text fieldPrime meridian as a >combo box (not editable for now)Semi-major axis lengthSemi-minor axis length >(later, we should add a button for switching to inverse flattening factor) >Coordinate system type (for now only Ellipsoidal or Cartesian)A table on two >or three rows containing the following columns:Axis name as a text fieldAxis >units of measurement >Axis direction (north, south, east, west) as a combo-box > > >If you can give me a detailed scenario of how you'll use the functionality you >are expecting in terms of user interface components then I can judge how much >work it will require. But I think that the tree table view can fit our needs >with the preferences functionality offering not only summaries but different >views. > >My proposal is: > >In the tree-table, limit the specialized widgets to those that fit on a single >line.Create a separated panel with no hierarchy and only the above-cited >metadata elements. More complex widgets appear there. > >Of course it can be questioned, and it is okay to try alternatives. > >Martin > >
