[cgma-dev] Inconsistency in getEntities for iGeom interface. Causes problems in iRel.

Tue Jan 27 09:23:32 CST 2015

OK, I found this old email thread.
Rajeev, we need to discuss what is expected with the attributes for rgg.
copy_attrib was enabled only for occ builds. So I assume that rgg with cubit already did what you want; my understanding is that copy attrib exposed a problem when rgg was built with occ.
after Jane fixed it, RGG worked fine with occ.
So we need to understand what is expected, and what attributes/tags are needed.

Attributes are treated differently for cubit and occ;

Iulian
________________________________
From: cgma-dev-bounces at mcs.anl.gov [cgma-dev-bounces at mcs.anl.gov] on behalf of Patrick Shriwise [shriwise at wisc.edu]
Sent: Wednesday, July 16, 2014 9:15 AM
To: Jain, Rajeev; Paul Wilson; CGMA Development
Subject: Re: [cgma-dev] Inconsistency in getEntities for iGeom interface. Causes problems in iRel.

Are there any default tags that are automatically generated upon creating a new entity in cgm? If so, should those be overwritten by the tags from the original entity we are copying?

Patrick

On 07/16/2014 02:35 AM, Rajeev Jain wrote:
I think all of the tags should get copied, but there might be use cases (in-future, I can't think of a case now) where making things more general and allowing a user option for selective or complete copying of tags might prove useful.

Rajeev

________________________________
From: Paul Wilson <wilsonp at engr.wisc.edu><mailto:wilsonp at engr.wisc.edu>
To: "Jain, Rajeev" <jain at mcs.anl.gov><mailto:jain at mcs.anl.gov>; shriwise <shriwise at wisc.edu><mailto:shriwise at wisc.edu>; CGMA Development <cgma-dev at mcs.anl.gov><mailto:cgma-dev at mcs.anl.gov>
Sent: Wednesday, July 16, 2014 12:09 AM
Subject: Re: [cgma-dev] Inconsistency in getEntities for iGeom interface. Causes problems in iRel.

Hi Rajeev,

Thanks for the response.  I guess it's still not clear what the expected behavior should be when copying an entity in iGeom.  Should it copy some/none/all of the tags to the new entity?  Is there a standard for what gets copied?

Paul

On 07/15/2014 08:26 PM, Rajeev Jain wrote:
Sorry, I missed this email.
Yes, you are right. It is for OCC based meshing. The test names volumes/surfs etc and then tries to retrieve them.
You can ignore the test case as we started working on developing other missing pieces for OCC geometry based meshing in MeshKit. When entities are copied the names can be copied with some an additional letter word under some convention.
Ex. CUBIT uses vol 1 named as mysteel when copied the new vol 2 is called mysteel at a etc.

Rajeev

________________________________
From: shriwise <shriwise at wisc.edu><mailto:shriwise at wisc.edu>
To: Paul Wilson <wilsonp at engr.wisc.edu><mailto:wilsonp at engr.wisc.edu>; "Jain, Rajeev" <jain at mcs.anl.gov><mailto:jain at mcs.anl.gov>; CGMA Development <cgma-dev at mcs.anl.gov><mailto:cgma-dev at mcs.anl.gov>
Sent: Tuesday, July 15, 2014 10:29 AM
Subject: Re: [cgma-dev] Inconsistency in getEntities for iGeom interface. Causes problems in iRel.

Bump.

On 07/11/2014 08:15 PM, Paul Wilson wrote:
Rajeev,

You are mentioned in the git logs as the inspiration for the copy_attrib test that Patrick mentioned.  (a few years ago)

I wonder if you have any thoughts on what the purpose and/or requirements are for this capability?  You probably needed it for RGG based on OCC?

Paul

On 07/11/2014 07:21 PM, Patrick Shriwise wrote:
Hi all,

I'm currently working on the changes to iGeom that Paul previously proposed. I've hit one test when built with OCC (copy_attrib) that suggests that tags are expected to be copied along with an entity when using iGeom functions like copyEnt or subtractEnt. Is this the case? Are these specifc tags that are supposed to be copied if they exist on the original entity?

Also to be clear this has become an issue because for volume entities we now use the RefVolume to get the volume Body and make a copy of that body using the GeometryModifyTool. The Body we make a copy of does not have the tags associated with the RefVolume. As a result the new Body and the corresponding RefVolumethat we return from the function does not have the original tags.

These tags can be transferred to the new entity before leaving the function of course, but we're looking for the proper way to handle this situation. Mainly it would be good to know if there is a particular tag or tags that should be copied over to new entity upon it's creation.

Cheers,

Patrick

On 06/23/2014 03:06 PM, Paul Wilson wrote:
Hi everyone,

Patrick and I have discussed this and have come up with the following proposal for a revision of how iGeom_CGMA handles RefVolumes and Body's.

  1.  All references to iBase_REGION will refer only to entities of type RefVolume in CGM.
  2.  All entities that are returned by iGeom will be RefEntities and not CubitEntities
  3.  At the time of loading file, based on an option based to iGeom_load(), all entities of type Body in CGM that contain more than a single volume, will be converted to RefGroup's and tagged to distinguish them from the original RefGroup's

The last of these may be subject to the most conversation, but it seemed like a useful approach to use because it:

  *   had little to no effect on the API of iGeom
  *   preserves access to CGM/ACIS bodies
  *   is the least intrusive in the implementation of iGeom_CGMA

If a user does NOT pass in the option (e.g. "CONVERT_BODY_TO_GROUP"), then no bodies will be accessible through iGeom.  If a user does pass in this option, any call to iGeom_getEntSets will return both the original RefGroups and the additional RefGroups that were generated automatically from Bodies.  It will be up to the user to filter between them using the available tag.
Finally, since every RefVolume in ACIS always has a Body in which it is a member, it seems best to only convert Bodies that contain >1 RefVolume to avoid creating far more RefGroups than necessary.
We'd like to embark on these changes soon, but welcome any comments or feedback.  I'm sure we've missed something!
Paul

On 06/17/2014 09:00 AM, Jane Hu wrote:
So, I will add RefVolume to the return list of iGeom_get_adjacent_entities() as the result of the discussion.

CubitEntity only has derived classes of RefEntity and CubitCoordinateSystem, iGeom_CGMA.cc only uses RefEntity as instance of the object.

Jane

On Mon, Jun 16, 2014 at 11:35 AM, Jane Hu <janejhu at gmail.com<mailto:janejhu at gmail.com>> wrote:
See comments below...

On Mon, Jun 16, 2014 at 10:22 AM, Paul Wilson <wilsonp at engr.wisc.edu<mailto:wilsonp at engr.wisc.edu>> wrote:
Hi Jane,

I think this needs a thorough review!  Patrick and I hunted through iGeom_CGMA and found some inconsistencies in a few places.

First, there is an inconsistency between returning pointers to RefEntity and pointers to CubitEntity.  This is probably less important, but should probably be resolved in any change.

I see RefEntity is inherited from CubitEntity, and CubitEntity is a pure virtual class. So even you got pointers to CubitEntity, you have to instantiate it to RefEntity objects.

Second, there are places where the geometry is queried for entities of dimension=3 (returning RefVolumes) and places where the geometry is queried for entities of type=iGeom_REGION (== Body).

In this case, I can return both bodies and refvolumes, so after user gets the refentity list back, he can cast to see if they are bodies or refvolumes, and choose to keep what ever he needs.

This means that 2 calls into the interface that should resolve to the same entity may not do so for both reasons.

All of this is related to the fundamental difference between the iGeom types (equivalent to vertex, edge, face, volume) and the CGM/ACIS types (equiavlent to vertex, edge, face, volume, body).  I think Tim's suggestion is the best, but will require some substantial effort.

Paul

On 06/16/2014 10:14 AM, Jane Hu wrote:
I see in iGeom_CGMA.cc file, function iGeom_get_adjacent_entities, if asks for 3-D refentities, it returns body list. In CGM, bodies are not necessary to be 3-D entities, it may contain single surface or shell. Should we change the return to be only refvolumes which are 3-D entities (I prefer this, makes more sense)? Or I can return both bodies and refvolumes in the returned refentity list. Depending on your user case, I can return one of the two choices.

Jane

On Mon, Jun 16, 2014 at 9:32 AM, Jane Hu <janejhu at gmail.com<mailto:janejhu at gmail.com>> wrote:
Yes, it returns a list of refVolume's which are RefEntities.

Jane

On Sun, Jun 15, 2014 at 11:55 PM, Patrick Shriwise <shriwise at wisc.edu<mailto:shriwise at wisc.edu>> wrote:
Hi Jane,

Will the ref_volumes() method return RefEntities?

Patrick

On Jun 15, 2014, at 3:56 PM, Jane Hu <janejhu at gmail.com<mailto:janejhu at gmail.com>> wrote:

Look at line 6454, if you have body's pointer, directly call ref_volumes() will work. It's not an iGeom interface function, but you can access it in iGeom.

Good luck!

Jane

On Sat, Jun 14, 2014 at 8:55 AM, Patrick Shriwise <shriwise at wisc.edu<mailto:shriwise at wisc.edu>> wrote:
Hi Jane,

I looked around for this but didn't see it, is there a specific line/function you were thinking of? Good to know the ref_volume information is so easy to get to, but it doesn't seem that its part of the iGeom interface. The issue being that there's no way to access a body's volumes via iGeom.

Cheers,

Patrick
On 06/13/2014 11:26 PM, Jane Hu wrote:

 Is there a way to get volumes within bodies via iGeom?

See example in  iGeom_CGMA.cc, body->ref_volumes(volumes).

Jane

--
Patrick C. Shriwise
Research Assistant
University of Wisconsin - Madison
Engineering Research Building - Rm. 428
1500 Engineering Drive
Madison, WI 53706
(608) 446-8173<UrlBlockedError.aspx>

--
Jane Hu