User:Xuwei/DevLog2019

Coummunity Bonding Period

• Set up the Development log
• Set up the development environment such as OS.
• Start to look at the different potential libraries code base.
• Discuss with mentors which external library to choose.

May 16th

• Start to check the checklist.
  • Agreement sent.
  • Dev log created.
  • My Profile created.

May 17th

• Had four of my wisdom teeth extracted. The progress for the following week may be slow due to teeth extraction.

May 19th

• Looking at Coin3D/Dime code base, a possible library for DXF import/export.
  • It has no external dependency.
  • It supports all past and should support the future version of DXF format.
  • It compiles on Windows, Linux, MacOS, etc.
  • But it only loads DXF files into Dime object hierarchies and saves Dime object hierarchies as files conforming to DXF format.
  • Seems only support 3D.

May 20th

• Looking at BRL-CAD DXF import/export code. Trying to understand how it works and see if it is suitable for OpenSCAD.
  • It converts DXF format to .g and .g to DXF which is a BRL-CAD format.
  • It seems that BRL-CAD does not support the entire format, only the part that is relative to solid-model.

May 21st

• Looking at dxflib which is used by QCAD and works fine, but it only supports 2D which is not ideal since OpenSCAD is already supporting 2D.
  • Maybe we can add the 3D part. But not sure how hard it is to add that.

May 22nd

• Searching for a library that supports both 3D and 2D DXF import/export.
  • Libraries above are for either 2D or 3D.
  • Found some libraries support both but they are written in python instead of C++.

May 23rd

• Continue on searching for the potential library.

May 24th

• Looking at and Compile libdxfrw code base which is used by LibreCAD.
• Will be out of town for the next two days.

Coding Period

May

May 27th

• Looking at both libdxfrw(libdxfrw.h, dxfreader.cpp) and OpenSCAD(dxfdata.cc, dxfdim.cc, import.cc) relative code.
  • Comparing the differences between these two on how they handle DXF data.
  • Trying to form a connection between these two.
  • dxfdata.cc seems a bit complicated. Need some time to digest.
  • The way that libdxfrw reads DXF file is kind of mixing with DRW code.
  • libdxfrw encounters problems when dealing with elevated arcs(no response yet).[1]
• Looking at dxflib code base as well.
  • dxflib does not store any entities, only pass supported entities.
  • A useful link for understanding dxflib data structure.[2]

May 28th

• Move on to reading dxflib code for now
  • Continue reading the link about the data structure of dxflib above.
  • Trying to understand form the connection between dxflib and OpenSCAD code base.
• Continue reading dxfdata.cc

May 29th

• Found another one existing integration with dxflib that I can use it as reference.CAMotics
• Continue on reading dxflib and import.cc, dxfdata.cc.
  • Keep trying to write some code that may help understanding how to integrate the library.
• Fill out the comparison wiki page

May 30th

• Had to go the dentist this morning, will catch up the afternoon and at night.
• Keep browsing for more information for those four libraries.
• Reading DXF standard to learn how DXF format works

May 31th

• Searching more information for potential libraries.
  • Add more information that found to the comparison page.
• Learning how to integrate the library from the instruction written by QCAD.

June

June 3rd

• We stick with using BRL-CAD's code for now.
• Try to implement the circle entity.
  • Add a callback function on dxf-g.c to pass the data back to openSCAD.
  • Add processentitiyecircle() to dxfdata.cc and some other changes to enable that function.
  • Having some issue with ADDLINE. Will dig into that later.

June 4th

• Push few commits on import_dxf branch.
  • Basically create a new import_dxf() function that replace the old dxfdata.cc
  • Implement a simple geometry to test out the function

June 5th

• Modify the code committed yesterday to output a simple circle.
• Trying to understand layer, block.

June 6th

• Created a new branch prototype on brlcadDXF repo.
  • Trying to clean up the code so that it could compile independently.

June 7th

• Switch some functions and datatype and continue on cleaning up the code.
• Spend most of my day studying the exam tonight. Will make up the time tomorrow.

June 8th

• Walked through the whole file line by line and cleaned up as much as I can so far. There are still many details that I need to learn to clean up the rest.

June 10th

• Made changes to dxf-g.c according to the answer from brlcad.

June 11th

• fix bu_list with std::list and related functions, etc.

June 12th

• Add functions for circle eclipse from vmath.h, color.h.

June 13th

• Added more functions from dependencies.
• uploaded a file that records to-do list, question, notes, etc.

June 14th

• few fixes on dxf-g.cc.
• Study for the midterm tonight for the most of the day will make up the hours tomorrow.

June 15th

• Finished converting all bu_list data structure to std::list except in one function drawmtext().
  • drawmtext() may be removed later.

June 17th

• Finished fixing all bu_vert_tree structure.
• Started on fixing vls a bit.

June 18th

• Finished fixing all vls except those in the main().

June 19th

• Finished most of the parts of dxf-g.c.
  • Create a new branch test for testing and commented out some less necessary functions to test.
  • Upload test files from openSCAD testdata/dxf
  • Need to fix the segmentation fault.

June 20th

• Fixed numbers of segmentation faults mainly caused by malloc.

June 21st

• Fix strncmp function error
• Tested with circle, ellipse, failed at block section.
• Tried to reconstruct block_list, block_head data structure.

June 24th

• Fix block_list data structure
• Now dxf-g.cc can go through the whole dxf file and spit out data to file output
• Done basic testing with circle, ellipse, polygons, lwpolyline. files from openscad testdata/dxf.
  • Found the correct geometry.
  • coordinates of the vertices are correct.

June 25th

• Randomly picking files from openscad/testdata/dxf to check geometry and coordinates.
• Trying to figure out if the data structure is actually storing the correct data.

June 26th

• Trying to extract data from the data structure to test if it's reading correct data.
  • For circle, extract cirlce_pts.
  • For line, extract line_pts.
  • For lwpolyline, extract polyline_vertice.
  • For elllipse, center and majorAxis
  • For point, pt.

June 27th

• More data extracting continue from yesterday.
• Try to start integrating openscad and dxf-g.cc just for circle entities.

June 28th

• Run openscad together with dxf-g.cc.
• Implemented circle entities using dxf-g.cc called a openscad function.
  • Later the day discussion concluded that this is not the preferred way. deprecated.

July

July 1st

• Added structs for each geometry for storing the data.
• Some minor fixes for original data type point_t.

July 2nd

• storing all the data parsed to the respect vectors.
  • Except polyline is a bit tricky, not done yet.

July 3rd

• Added polyline struct and the storing part.
• Added code for passing circle entity information to openscad and openscad draw it.
  • Having trouble with compilation.

July 4th

• Added arc, line, points, lwpolyline.
  • Still have some problems with lwpolyline.

July 5th

• fix lwpolyline.
• Added ellipse.
• Found problem with mesh.

July 8th

• Trying to fix mesh

July 9th

• Change the structure of the entire code to use the code of dxfdata to fix the mesh problem
• The code still not able to produce the correct polygon yet.

July 10th

• Fix the mesh problem and now the code can properly import a single dxf file.
  • For some reason, the data is store and fill in the next polygon if the second dxf is imported.

July 11th

• Fix the data leakage between document.
• Now the code is passing all the test file except nothing decimal comma and transform insert.
  • the reader has an issue with insert entity

July 12th

• Trying to figure out how to fix the insert.
• Learning the insert spec from dxf manual

July 14th

• Fix the block list and curr_block is not storing value properly
• There is an infinite loop caused by the file offset in the process_enitites_unknown
  • It jumps to the previous line but does not jump back
• Continue on fixing insert

July 15th

• Clean up most of the commented code in dxf.cc and dxf.h
• fix the curr_block memory issue by using a global variable indx that indicate the position of the block in block_list
• Continue on fixing insert

July 16th

• Adding back_file_offset to state_data to have the program jump back
  • Still getting the wrong result, the geometries are scattered but shapes are the same.
• Continue on fixing insert

July 17th

• replace all memory allocation function related to insert with new and delete
  • valgrind now output no error or warning
• Continue on fixing insert

July 18th

• Download and build brlcad to test out what's the difference and if it can output the correct result
  • There is no problem with the curr_state->file_offset
• Continue on fixing insert

July 19th

• Found that the brlcad xform is different than my version.
  • Can't figure out why it's different
• Continue on fixing insert

July 20th

• Found the BU_LIST_POP actually assign curr_state to the last element of the list
  • This fix the infinity loop and remove back_file_offset.
• Continue on fixing insert

July 21st

• By switching the push_back and new_state = curr_state this fix the xform problem.
  • Insert is fixed but it looks there is an issue with different compiler

July 22nd

• Cleaning up dxf.cc, import-dxf.cc, dxf.h.

July 23rd

• Finished cleaning up dxf.cc, import-dxf.cc, dxf.h.
• Make #pr3005 to test in various platform.

July 24th

• Fix scale, xyorigin, layername problem.
• Test result [3]

July 25th

• Starting working on spline entity.
• Reading spline related code on QCAD.
• Learning dxf spline entity spec.

July 26th

• QCAD is using an external library opennurbs to deal with the spline.
• Reading spline related code on BRL-CAD.
  • BRL-CAD code can be implemented on openscad.

July 29th

• Replicate and integrate brlcad spline code to openscad.

July 30th

• Continue integrating blrcad spline code to openscad.

July 31st

• Testing brlcad code with openscad, comparing with brlcad and qcad.
  • Creating test cases using AutoCAD.
  • Both brlcad and qcad seems to not supporting the spline created by autocad properly due to dxf version problem.
  • brlcad encounter segmentation fault with one of the spline test cases.

August

August 1st

• Switch to reading LibreCAD spline related code.
  • LibreCAD 2.x version is using their own code.
• Also reading openscad bezier curve related code.

August 2nd

• LibreCAD can handle the spline created by autocad unlike brlcad and qcad.
• OpenSCAD code can handle degree 2, 3 normal spline(Beizer curve) properly.
  • Starting adding openscad code to import_dxf.

August 3rd

• Modified openscad code and integrated it to import_dxf.

August 4th

• Testing the spline import of openscad
  • Able to import 3 control points for degree 2 spline and 4 for degree 3 spline.
  • Fail to generate continuous bezier curve.

August 5th

• Trying to figure an algorithm or a way to generate continuous bezier curve like this article.

August 6th

• OpenSCAD is now able to import degree 1-3 spline with certain numbers of control points.
  • spline knots, weights, fit points are ignored.

August 7th - 14th

• Family trip.
• Originally from 7th-12th, due to the riots in Hong Kong airport and the typhoon in Osaka, flight got cancelled and delayed. ** Due to the typhoon, updated flight time is 16th.

August 15th

• Starting working on polyline entity.

August 16th

• Rearrange code and fix error in storing polyline vertex in the importer.
• On the flight home

August 17th

• Integrate 2D polyline entity to openscad ignoring 3D polyline features and curve-fit.

August 18th

• Improve polyline with the spline-fit flag, can generate by both spline vertex and control points.
• Test polyline.
• Add test cases to the test suite.