In the file dialog and will click on it without knowing what it does and ends up in a strange situation with timesteps that do not make sense and missing data.
But the original timeseries is also great !
A user can just create multiple files from a simple simulation code and see the results animated in ParaView in a matter of seconds.
Finally, grouping multiple files together can be complex as some file formats just do not support it and doing it manually in ParaView may require lots of clicks when dealing with lots of datasets.
What about letting the user choose and expliciting the concept in the meantime ?
When opening a series of file : file....ext, show a dialog that let user choose if they want a timeseries or a groupseries, and explaining what it means in the process.
Less lost user, a single supllementary click for timeseries user and less click for the user of the new groupseries features.
We could imagine the same concept, but done explicitely, in the .series files.
I’m trying to understand the purpose of the groupseries. Is it to replace reading in a bunch of separate files and then use the Group Dataset filter to group them into a composite dataset?
If that’s the use case then what about the situation for when the datasets have different file extensions?
Are you mainly concerned about the case of files named like part1.vtk, part2.vtk, part3.vtk, … where the files would appear like part..vtk in the file browser and cause problems when reading in to ParaView if the user didn’t expand them out?
If that’s the use case then what about the situation for when the datasets have different file extensions?
I dont think it should be supported but it could.
Are you mainly concerned about the case of files named like part1.vtk, part2.vtk, part3.vtk, … where the files would appear like part..vtk in the file browser and cause problems when reading in to ParaView if the user didn’t expand them out?
Yes, but also add a groupseries features to easily grouping datasets together.
Ah now that I think about it more, I see that it would make sense for (some) other formats as well.
I suppose for formats that don’t otherwise have an internal method for specifying whether it’s part of a timeseries or fileseries. I’m thinking of simple formats like PLY or STL.
More complex formats like ExodusII wouldn’t need these additional options since the data format/reader already handles these distinctions, right?
Indeed, formats that already supports timesteps do not need fileseries. However, regarding groupseries, it could even be added on any format, as adding a composite data layer is always possible.