1. 10 Nov, 2019 1 commit
  2. 09 Nov, 2019 1 commit
  3. 19 Jul, 2019 2 commits
    • Scott Kilpatrick's avatar
      Now properly defining the ANALYZED set, and fully analyzing that and SELECTED sets too. · c5f5e4f7
      Scott Kilpatrick authored
      Previously I seem not to have properly run the analysis on the entire SELECTED set. Additionally, I was not taking the full, transitive dependency closure of SELECTED in order to derive ANALYZED. That has been corrected.
      c5f5e4f7
    • Scott Kilpatrick's avatar
      Major reproduction of the results · 6aaeb2bd
      Scott Kilpatrick authored
      After revisiting the entire analysis, this time with more scrutiny
      for the final thesis write-up, there were some changes in the results.
      Notably, there are not merely 10 inconsistent packages from the
      `selected-popular` category but actually 21 of them. The data files
      have also been rearranged with a more consistent style.
      
      Additionally, a new program, `bin/original-sin.py`, is useful for counting
      occurrences of distinct `package:module` references in text files.
      6aaeb2bd
  4. 17 Jul, 2019 1 commit
  5. 27 Jul, 2018 5 commits
  6. 23 Jul, 2018 2 commits
  7. 22 Jul, 2018 2 commits
    • Scott Kilpatrick's avatar
      Some shameless attempts to force evaluation · d7720bf7
      Scott Kilpatrick authored
      Also removing ModIface from CtxEntry so that we don't hold onto more
      references than we need.
      d7720bf7
    • Scott Kilpatrick's avatar
      Major rewrite of worlds to include caching · 5f12f95f
      Scott Kilpatrick authored
      Now all world mergeability will produce inconsistency reports in the
      form of clashing Islands. Most of them will also be checked in a monad
      that keeps up a cache of world consistency. That cache tries to remember
      all `(m1, m2)` pairs such that any instances known to `m1` are mergeable
      with any instances known to `m2`. This cache gets huge, so I need
      heuristics to keep its size down.
      5f12f95f
  8. 21 Jul, 2018 1 commit
  9. 02 Jul, 2018 1 commit
  10. 22 May, 2015 1 commit
  11. 21 May, 2015 6 commits
  12. 20 May, 2015 6 commits
  13. 18 May, 2015 1 commit
  14. 17 May, 2015 4 commits
  15. 16 May, 2015 1 commit
  16. 15 May, 2015 1 commit
    • Scott Kilpatrick's avatar
      Executables now take package lists from STDIN; adding orphan results · 4d04377c
      Scott Kilpatrick authored
      Now the calc-worlds and orphans-stats executables take in the list of
      packages to analyze from STDIN. The input format assumes each line is a
      package name (with no extra whitespace), an empty line, or a comment
      that begins with '#'.
      
      Added the results of orphans-stats on platform packages, popular
      packages, and all combined.
      4d04377c
  17. 13 May, 2015 4 commits