ParaView 3 Telecon 04-08-2009: Difference between revisions
| (One intermediate revision by the same user not shown) | |||
| Line 34: | Line 34: | ||
== pvserver/pvws/tws consolidation == | == pvserver/pvws/tws consolidation == | ||
The Titan Web Services (TWS) and ParaView Web Services (PVWS) are currently completely separate projects. Since there are people interested in using both at the same time, | The Titan Web Services (TWS) and ParaView Web Services (PVWS) are currently completely separate projects. Since there are people interested in using both at the same time, it becomes necessary (or at least stupid not to) consolidate these two. | ||
The second point was that there is some interest in providing paraview web services on HPC systems when doing, for example, in situ. However, there is a barrier to entry of doing that in that you must first set up a PVWS before getting anything to work. So, ''what if'' the ParaView client could connect through the PVWS instead of directly via ssh scripts and sockets? That could reduce the pain of setting up a ParaView service on an HPC system. But it would also provide a ready to use PVWS that could also be used for a variety of other types of web clients. | |||
This cannot be done yet. The current web interface is much smaller than the general pvserver interface. That said, Utkarsh has a task to think about what it would look like to have these together and if this is feasible. | |||
== Label vs. Name in SMXML == | == Label vs. Name in SMXML == | ||
| Line 41: | Line 45: | ||
I thought we informally agreed that all labels should be basically the same as the name with explicit spacing unless there was a good reason to do otherwise. We even talked about enforcing this in the XML parsing. However, I am still seeing new proxies and properties being added that break this rule. (I most recently noticed the AMRDualClip/Contour proxies.) | I thought we informally agreed that all labels should be basically the same as the name with explicit spacing unless there was a good reason to do otherwise. We even talked about enforcing this in the XML parsing. However, I am still seeing new proxies and properties being added that break this rule. (I most recently noticed the AMRDualClip/Contour proxies.) | ||
Action Item: Add to the bug tracker a request to check names and labels to be the same. It will likely have a low priority because we are not sure if we really need something like this. | |||
Latest revision as of 12:56, 8 April 2010
| Item | People | Description |
|---|---|---|
| 1 | Moreland | Project Randy |
| 2 | Utkarsh | 3.8 Release |
| 3 | Utkarsh | Transition to git |
| 4 | Moreland | pvserver/pvws/tws consolidation |
| 5 | Moreland | Label vs. Name in SMXML |
| 6 | Moreland | Stronger Dashboard
|
Project Randy
Image rendering saving seems to be working.
3.8 Release
First release candidate eminent.
Transition to git
On the week of the 19th, there will be a code freeze to transition to git.
pvserver/pvws/tws consolidation
The Titan Web Services (TWS) and ParaView Web Services (PVWS) are currently completely separate projects. Since there are people interested in using both at the same time, it becomes necessary (or at least stupid not to) consolidate these two.
The second point was that there is some interest in providing paraview web services on HPC systems when doing, for example, in situ. However, there is a barrier to entry of doing that in that you must first set up a PVWS before getting anything to work. So, what if the ParaView client could connect through the PVWS instead of directly via ssh scripts and sockets? That could reduce the pain of setting up a ParaView service on an HPC system. But it would also provide a ready to use PVWS that could also be used for a variety of other types of web clients.
This cannot be done yet. The current web interface is much smaller than the general pvserver interface. That said, Utkarsh has a task to think about what it would look like to have these together and if this is feasible.
Label vs. Name in SMXML
Not long ago we had some problems with Python scripting where many names of proxies and properties were significantly different from their label. This cause confusion, especially when transitioning between the GUI, Python scripting, and C++ code.
I thought we informally agreed that all labels should be basically the same as the name with explicit spacing unless there was a good reason to do otherwise. We even talked about enforcing this in the XML parsing. However, I am still seeing new proxies and properties being added that break this rule. (I most recently noticed the AMRDualClip/Contour proxies.)
Action Item: Add to the bug tracker a request to check names and labels to be the same. It will likely have a low priority because we are not sure if we really need something like this.
