User:Vikram Atreya/GSoC21/Log

From BRL-CAD
< User:Vikram Atreya
Revision as of 15:01, 5 July 2021 by Vikram Atreya (talk | contribs) (Week 4 update)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Development Logs[edit]

Community Bonding Period[edit]

  • Forked github repo and compiled working version of BRL-CAD
  • Made the pages for project plan and dev logs
  • Ideated various ways to implement undo and set constraints

Weekly update - Coding Period[edit]

  • Week - 1
    • Day 1-2
      • Working version of undo command implemented in mged which prints name of database when command is given.
      • Understood how commands use wrapper functions and function of cmd_ged_plain_wrapper()
      • Set flags -n and -h aimed to execute multiple undos and show usage respectively
      • Understood how hidden objects work, have to explore temporary file alternative to save backup objects
      • Stuck, trying to iterate over all objects in database but not finding the right function
      • Finally found way to iterate over all objects from move_all.c
      • Converted undo.c to undo.cpp so that code can be more functional
    • Day 3-4
      • Had a discussion with my mentor how to proceed further on the undo
      • Data accesible over all processes stored in _GLOBAL as name, value pairs
      • Name of the last added object and last killed object stored in _GLOBAL
      • Wrappers of make and kill modified to add the labels
      • Introduced 3 variables in undo.cpp to keep track of latest_change and then take necessary action to undo
    • Day 5-6
      • Enabled undo for in command; Now working to enable undo for killall
      • Did tutorials to understand how combinations and regions work
      • Went in the wrong direction intially; Took a day to understand how combinations are stored
  • Week - 2
    • Day 1-2
      • Encountered many bugs while implementing undo for killall
      • Ubuntu had some problems; mouse and keyboard stopped working and stopped work for sometime
      • Tired refactoring code of backup storage for make; Chose to make a gmd_cmd for it but did not happen, also felt it was not the best option
      • Succesfully implemented undo for kill by modifying code in libged/killrefs; Might have to write a new wrapper for this
      • Created new wrapper for g (group) command and implemeted undo for it; Some refinement left here
    • Day 3-4
      • Refined code of g cmd undo and updated PR
      • Attempt at reefactoring code to libged/undo.cpp failed; Not right place to put the code
      • Refactored make's undo code into mged/utility1.c and replaced code in make, in and g
      • Solved major bug that made children objects disappear when any object was killed
      • logic: renamed all leaves by adding sufffix, thereby poiting to imaginary objects and not hiding the references it had previously
    • Day 5-6
      • Implemented undo for mv and mvall
      • Also restructured code in undo.cpp to accommodate more cmds easily
      • Made a spreadsheet with status of implementation for all commands
  • Week - 3
    • Day 1-2
      • Solved a long pending bug
    • Day 3-4
      • Took a break
    • Day 5-6
      • Started work on storing objects as data in binary objects
      • Understood code of mk_binunif, make and libged/bo.c
  • Week - 4
    • Day 1-2
      • Done writing basic test to store and recover objects from a binary object
      • Stored internal data in binary object, storing external data is better since it contains type information as well
      • Switched code from storing internal to external of object
    • Day 3-4
      • Ideated and tried different ways in which we could store the recovery data
      • Wrote code implementing a new struct in ged/defines.h
      • Mimicked history implementation for the struct
    • Day 5-6
      • Implemented undo using the struct but realized it would not work
      • Thought of using char** arrays to store the commands
      • Have to discuss with mentor on a better approach