support variables with empty/missing blocks apart from underlying mesh

[aowen87]

This is requested by LLNL Engineering (not GrizIt team though) and favored by LLNL WCI. Currently, if a multiblock variable is defined on only some blocks, the data producer is required (by VisIt not by Silo) to create a cooresponding multiblock mesh object with the same missing blocks. This cause the data producer to create mesh objects representing various subset configurations that are otherwise easily defined/constructed by SIL restrictions. In the new approach, a data producer will be allowed to create a single multiblock mesh object and then define upon it multivar objects that have missing/empty blocks. We should decide on best approach for how VisIt will then treat a missing variable block. Some options considered so far are... Missing variable blocks get filled with a fill value (where data producer and visit user can select fill value) Missing variable blocks are treated essentially as transparent (ghosted away) Missing variable blocks are colored specially (e.g. kinda like greyed out where data producer or visit user can select color) This approach requires some color be set aside from the normal colormaps that can easily distinguish the block from normal data. Missing blocks impose a modified SIL restriction such that the missing blocks are turned off in the SIL and also grey'd out in the SIL controls window. Works fine when SIL is different for each plot (e.g. not using Apply SIL to all plots option) I think it may make sense to support all of these appraoches in VisIt's interface(s) so that users can vary behavior as needed.

-----------------------REDMINE MIGRATION----------------------- This ticket was migrated from Redmine. As such, not all information was able to be captured in the transition. Below is a complete record of the original redmine ticket.

Ticket number: 3051 Status: Pending Project: VisIt Tracker: Feature Priority: High Subject: support variables with empty/missing blocks apart from underlying mesh Assigned to: Mark Miller Category: - Target version: 3.0 Author: Mark Miller Start: 03/28/2018 Due date: % Done: 0% Estimated time: Created: 03/28/2018 01:06 pm Updated: 05/10/2018 01:05 pm Likelihood: Severity: Found in version: 2.12.3 Impact: 3 - Medium Expected Use: 3 - Occasional OS: All Support Group: Any Description: This is requested by LLNL Engineering (not GrizIt team though) and favored by LLNL WCI. Currently, if a multiblock variable is defined on only some blocks, the data producer is required (by VisIt not by Silo) to create a cooresponding multiblock mesh object with the same missing blocks. This cause the data producer to create mesh objects representing various subset configurations that are otherwise easily defined/constructed by SIL restrictions. In the new approach, a data producer will be allowed to create a single multiblock mesh object and then define upon it multivar objects that have missing/empty blocks. We should decide on best approach for how VisIt will then treat a missing variable block. Some options considered so far are... Missing variable blocks get filled with a fill value (where data producer and visit user can select fill value) Missing variable blocks are treated essentially as transparent (ghosted away) Missing variable blocks are colored specially (e.g. kinda like greyed out where data producer or visit user can select color) This approach requires some color be set aside from the normal colormaps that can easily distinguish the block from normal data. Missing blocks impose a modified SIL restriction such that the missing blocks are turned off in the SIL and also grey'd out in the SIL controls window. Works fine when SIL is different for each plot (e.g. not using Apply SIL to all plots option) I think it may make sense to support all of these appraoches in VisIt's interface(s) so that users can vary behavior as needed.

Comments: From an engineering user...So, the analysts over here often build models with lots of little components/material definitions. It's not too unbelievable that a model could have only 23 very small parts that have a particular field defined. For the SIL restriction case, this could result in 9899% of the model being disabled, which would remove any possible context which would make discrimination of these multiple little parts possible. One can imagine a case where these little parts have significantly different responses, but one can't really tell if it is, say, the "left" or "right" instance of the part that has a "problem", because too much of the remainder of the model is not visible.Of course, one might also imagine cases where this is not the case . . . I am not saying that SIL restriction is never useful, only that there are enough cases where it would be supremely unuseful, that it can't be the only option. So, if it is one option among several options, cool. From a WSC code developerI would personally expect it to be ghosted away. I think using a fillvalue could be misleading, but it might be best to get a couplepeople's opinion. Having an option on how to treat it in preferencesmight also be a way to try to please everyone.