---
tags: decompiler
title: Timeline
---

# Overview

## Dashboard:
- [x] Complete rule table [name=Ali] - **Saturday Midnight**
- [x] Categorize them
- [x] Add in paper :warning: In process
- [x] Add more implicit errors
- [x] Find implicit error functions [name=Chijung]
- [x] Inspect for unique new implicit errors in 2000 remaining samples [name=Chijung]
- [x] Centralize and add them in the paper [name=Ali]
- [ ] (V-C) Complete Eval [name=Ali] - **Sunday Midnight**
- [x] Most Effective Transformation Rules
- [x] Explain top 3 rules
- [x] Fill XXX
- [x] # of Transformations Applied
- [x] Report rules applied and number of rules applied per block
- [x] Errors per File/Function
- [x] Count of files and functions with 2 or more errors
- [x] Most number of errors in file
- [x] Most number of errors in single function
- [x] Types of Transformation Rules Applied
- [x] Explain top rules for each python version
- [x] Comment on implicit errors
- [x] Concrete numbers of rules for Python 2.7
- [ ] (V-D) Time overhead calculation [name=Ali] - **Monday 6pm**
- [ ] Craft Experiments - **Sturday Midnight**
- [ ] Time taken for applying rule w.r.t file size
- [ ] Time taken for decompilation w.r.t file size
- [ ] Add results in paper
- [ ] Handling Opcode Remapping Case study [name=Ali] - **10pm Monday**
- [ ] Recheck
- [ ] Cover more remaining samples [name=Ali]
- [ ] Resolve more errors
- [ ] Update Eval table to reflect coverage

## :alarm_clock: Timeline
- Saturday April 2:
- Deadline @ 7:59 am

## :closed_book:Overall tasks
- [x] 1. Case studies [link](https://hackmd.io/@aliahad97/rkm8HC9Z9):
- [x] a. Python 3.9 case study
- [x] b. Regular case study of malicious sample
- [x] c. Customized python vm - shuffled instructions
- [ ] 2. Evaluation section [link](https://hackmd.io/@aliahad97/ByCo2tFb9)
- [x] a. Samples statistics
- [x] b. Error inducing statistics
- [x] c. List down what can and cannot be
- [x] d. Add initial tables to paper
- [x] e. Find stats related to blocks
- [ ] f. Evaluate correctness
- [ ] g. Evaluate time taken, efficiency and performanc
- [ ] 3. Cover python 3.9 cases ([rules](https://hackmd.io/@aliahad97/HkDD5eVWq), [progress](https://docs.google.com/spreadsheets/d/10dA4An1F36qm5aruGctzKn6nRIQk8xQtvaYodVtsoMI/edit#gid=1627318841))
- [x] a. Initial draft
- [x] b. Add to Paper
- [x] c. Finalize rules
- [x] d. Finalize transforamtions
- [ ] e. Reflect changes in paper
- [x] 4. Handle "Other errors" [link](https://docs.google.com/spreadsheets/d/1i3dRGD0GWnQ9OlN7ajnxxfHSSFn5N_yVIgcSytjc3CA/edit#gid=0)
- [x] a. Check up on `parse errors`
- [x] b. Inspect other errors
- [x] c. Finalize draft and charts for these errors
- [x] 5. Eval CFG changes "After" decompilation [name=Chijung]
- [x] a. Compare CFG in mal dataset
- [x] b. Compare CFG in benign dataset (ground truth)
- [ ] 6. Update dataset numbers [link](https://docs.google.com/spreadsheets/d/1lWiTob6nIFrQFSZFpIHcUmtopqbEJNi0JVm1GPklqTQ/edit?pli=1#gid=1429326896)
- [x] a. Prune library related samples
- [x] b. Merge Malware with pyinstaller samples
- [x] c. Create table for resolved errors
- [ ] d. Resolve that cannot be resolved and reaason if cannot
- [ ] 7. Paper tasks
- [ ] a. Discussion section
- [ ] b. Add centralized rules
- [ ] 8. CFG
- [ ] a. Complete the entire list of implicit errors

# :book: Notes:
- Percentages should be around 5%
- Don't rely on percentages
-

# :handshake::pencil: Meeting notes:

## Meeting 2
- Use diagrams
- Py3.9
- Are these big changes?
- Case study
- shouldn't be trivial
- failing point
- Stitch mulitple samples
- Go visualize later
- Create tables for now
- Eval:
- Basic blocks
- run regex and see match
- CFG done by:
- Clear picture By monday!
-
## Meeting 1
- Examples of changing CFG
- Why this happens?
- Versions diff
- recompilation test with benign source code samples
- Case studies
- py3.9
- customized python vm - shuffled instructions
-
- Find numbers
- What we have
- What we can attain
- How many rules
- Basic blocks transformed
- How many errors
- iterations for a function and iteration for a file
- Correctness?????
- number of errors
- average of errors
- Samples source of when and where we got them
- Transformation lead to new error and what we do about those?
- Push all to 100% - remove library files
- Prepare for other errors

## IF ASKED
- Does priority of rule impact the tool
- priority can be devised for performance but it doesn't affect the validity