martinezmatias / diff-auto-tuning Goto Github PK

cvs2git --blobfile="blob_carol.dat" --dumpfile="dump_carol.dat" /Users/matias/develop/gt-tuning/git-code-gpgt/results_finals/cvs-vintage/carol/ --retain-conflicting-attic-files --encoding="latin1"
git init
git fast-import --export-marks=git-marks-carol.dat < blob_carol.dat
git fast-import --import-marks=git-marks-carol.dat < dump_carol.dat
git checkout

https://www.mcs.anl.gov/~jacob/cvs2svn/cvs2git.html
https://alathurne.wordpress.com/2015/07/26/migrate-my-old-cvs-to-github-tools-and-problems/

TODO LIST:

• Using API of GT3 instead of GT.Spoon3
• Define a Java API: results must be stored in CSV (as done in v1) but it should be also returned, so they can be used via Java API
• Decoupling from MegaDiff (Main and Tuning Engine)
• Move to Java implementation of TPE (It was previously done in Python, which read the outputs from JavaDat i.e., the CSV)
• Define architecture of new experiment: before the summarization was done in Python
• Check parallelisation
• Update GT3 (pull, and install)
• check cases where JDT produced a strange value. Check in GT command line
• Parse space specification from XML
• Store the edit script
• edit API for calling other algos
• Add simplified Edit Script?
• Prepare CVSVintage
• Move to GumTreeSpoon2, as we need only the Builder/Parser
• SingleDiffResult is still a map: move to a pojo. The diff is not still stored in the map.
• Check save diff in both TPE and Exhaustive
• Create mode random from TPE
• Create interface to call Exhaustive and analyze the results
• Constant class to enum
• Control number of threats
• Put Proxy on engine as constructor parameter.
• Create a clear interface for GTproxy
• See paralelisation and TPE
• Save edit scripts
• Try old version of GumTree to check paralelism
• Check results on grid are fine for doing Global/Local analysis