User:Bhollister/DevLogJune2015

---

Monday, June 1, 2015: Start of Week 2 (of 14)

• dissertation work

Tuesday, June 2, 2015

• dissertation work

Wednesday, June 3, 2015

• dissertation defense @ 12:45-2pm PDT

Thursday, June 4, 2015

• added callable nmg_mrsv stub, see: https://github.com/behollis/brlcad-svn-rev65072-gsoc2015/commit/ff812753027d4119a590d9d0096f9dbb279163b0
• successfully defended PhD dissertation yesterday
• minor additions / editing / proof-reading of dissertation before ProQuest submittal deadline of June 11th
• todo: add remaining CONSTRUCTION routine stubs
• todo: determine how to pass nmg model verts, etc. from command line (mged) to function stubs

Friday, June 5, 2015

• received a few pointers on NMG, and a document link in weekly mtg
• dissertation editing

Saturday, June 6, 2015

• dissertation editing

Sunday, June 7, 2015

• dissertation editing

Monday, June 8, 2015: Start of Week 3 (of 14)

• commit #1 for today: added remaining NMG construction command stubs
• commit #2 for today: nmg_mrsv command now accepts a model file name and checks the database; removed nmg_mm since make cmd covers that functionality
• commit #3 for today: fixed format spacing for nmg_mrsv.c

Tuesday, June 9, 2015

• final dissertation edits
• submitted final dissertation to ProQuest
• delivered signatures, etc. for degree
• took online surveys for degree
• note #1: (todo) need to talk to mentors about cmd line interface; some construction routines in "Combinatorial Solid Geometry, Boundary Representations, and Non-Manifold Geometry," are likely too low-level. for instance, does it make sense to specify 'vertexuse' / 'edgeuse' at the cmd line?? possibly only routines that require user specified *new* vertices and edges will be implemented... also, we may not want to implement some of the api functions that only create empty shells / regions for new models. models can already be created using the cmd make.
• note #2: construction routines don't require passing particular instances of existing geometry. this makes their impl likely easier than destruction routines in terms of specifying geometry at the CLI.

Wednesday, June 10, 2015

• commit #1 for today: added call to internal api for "Make Region, Shell, Vertex"
• tried various testing with creating nmg geometry, but was not fully able to create nmg test cases with primitives. need to retry using "facetize" command again or another route
• need to determine if we need a higher-level CLI than internal NMG access API functions; some of this is already implemented in nmg_simplify, nmg_fix_normals, make, etc.

Thursday, June 11, 2015

• commit #1 for today: added impl for nmg_mm; small fix for return code from nmg_mrsv CLI libged function
• friday todo: implement nmg_mmr()
• friday todo: implment nmg_msv() - figure out how to specify region and pass that to CLI from MGED display (or archer)

Friday, June 12, 2015

• need to attend graduation ceremonies practice for a few hours (starts @ 1 PM PDT, but need to pick up gown, etc.)
• commit #1 for today: added impl for nmg_mmr for CLI. almost identical to ged_nmg_mm, but region is created for the new model.
• commit #2 for today: added preliminary nmg_msv CLI command. command accepts model name and adds shell / vertex to the first region encountered via BU_LIST_FIRST. the command also performs NMG_CK_REGION. if a model doesn't have a region, then this check causes MGED to log a bad pointer error and crash. TODO: may need to find a way to specify region to CLI and handle error w/o crash.

Saturday, June 13, 2015

• relatives visiting thru Thursday

Sunday, June 14, 2015

• graduation ceremonies attendance (most of the day)

---

---

Monday, June 15, 2015: Start of Week 4 (of 14)

• added nmg_me imp for CLI. not committed to github yet due to merge conflicts?? todo: fix this tomorrow!
• read thru some-thoughts-about-nmg-primitve
• concerning some-thoughts-about-nmg-primitve, not sure if shells or regions will be specified or used on CLI.
• todo: need to find problem with nmg_msv CLI. appears that it is not adding new shell to nmg data structure
• todo: fix nmg_me CLI to allow for setting vertex coords (current impl only adds vertexuses via nmg_mvvu)

Tuesday, June 16, 2015

• commit #1 yesterday: added nmg_me to CLI. can't currently set vertex coord but the vertexuses are added properly.
• commit #1 for today: adding CLI version of nmg_me. command takes two vertex coords and the name of the model. the first region / shell are used. also, updating comment in ged_nmg_msv.
• commit #2 for today: added ged_nmg_mvvu. routine takes first region / shell as parent structure in model to add vertex / vertexuse. note that a NULL pointer is added to the vertexuse struct. although possibly not clear, ged_nmg_mvu can be used to set a specific vertex instance for the vertexuse, i.e. a *new* vertexuse.
• commit #3 for today: ged_nmg_mvu impl prelim. need to find vertex with coordinates given on CLI. Use the parent struct of the found vertex as upptr. Here, this is a shell. TODO: the point of this routine is to re-assign a vertexuse to an *existing* vertex. Right now, this routine finds a shell with the specified vertex and replaces the vertexuse in that shell. we need to find a vertexuse *not* in a shell, such as an edge, and reassign it a *new* vertexuse. CLI doesn't use vertex id, so coord may be non-unique to multiple vertices in nmg model. IMPORTANTLY: crashing on curr_vg = curr_s->vu_p->v_p->vg_p; Need to fix.
• todo: need to fix nasty white spacing issues in submitted code revisions!

Wednesday, June 17, 2015

• last day of relatives' visit. taking day off. will work some saturday as makeup.

Thursday, June 18, 2015

• commit #1 for today: added ged_nmg_ml. calls nmg_ml directly with the first encountered region / shell in model. if shell doesn't have the required edgeuses / edges, then program throws and error.
• commit #2 for today: added ged_nmg_meonvu. user can specify a vertex coord on CLI. routine searches for the first encountered vertexuse that uses a vertex instance (coord). note: this is not what the internal api function is meant to do. it uses a *vertexuse*. todo: pass a *vertexuse* if possible from the CLI, or find another way to define similar construction routine that is more suitable for the CLI. once the vertexuse / vertex / geom is found, an "empty" edge with the same topological vertex(use?) is inserted into the model.
• found how to properly traverse model with BU_LIST_PNEXT macro. todo: need to update ged_nmg_mvu to reflect this correction.