1. 26 Oct, 2014 2 commits
    • Carlos Garcia Campos's avatar
      regtest: Limit the number of worker threads to the number of documents to test · 68e58ec5
      Carlos Garcia Campos authored
      We are always spawning all the threads even if the documents to test is
      less than the worker threads. Also optimize the case of running only one
      test to not spwn any thread.
    • Carlos Garcia Campos's avatar
      regtest: Allow to run groups of tests individually · 7c21b958
      Carlos Garcia Campos authored
      Now it's possible to pass more than one argument to run-tests command and
      optionally the docs directory. When more than one test is passed and the
      docs directory is not provided, the common base path of all passed tests
      is used as docs directory. The tests passed can be documents or
      directories, using absolute paths or paths relative to the docs
      This also allows us to update the refs for a group of tests.
  2. 15 Dec, 2013 2 commits
  3. 06 Dec, 2013 4 commits
  4. 29 Nov, 2013 4 commits
    • Carlos Garcia Campos's avatar
      regtest: Do not buffer stderr output · 817cc333
      Carlos Garcia Campos authored
      Some buggy documents can produce a huge stderr output because of parsing
      errors or whatever. We could give a file directly to Popen to write the
      stderr file, but we only want to create the file when there's output,
      because it's what we use to know whether the command produced output or
      not. So, instead of buffering the whole output and then write it to the
      file, now we read from the pipe while the command is running, writing
      the output in chunks to the file. This improves a lot the memory
      consumption when running some tests.
    • Carlos Garcia Campos's avatar
      regtest: Read test results in chunks to get the md5 digest · f8f82f1c
      Carlos Garcia Campos authored
      Some backends can generate huge results, like huge postscript files that
      we don't want to load in memory to get the md5. So, instead of creating
      thr md5 object with the entire file, we feed it with chunks of data
      using the update method. This improves a lot the memory consumption and
      performance as well.
    • Carlos Garcia Campos's avatar
      regtest: Remove unused method _check_exit_status2 · 3444a443
      Carlos Garcia Campos authored
      It was used when the backends ran in parallel odd and even pages, but
      it's no longer used since threads support was added.
    • Carlos Garcia Campos's avatar
      regtest: Do not store the current line in Printer but only its length · 5f825df4
      Carlos Garcia Campos authored
      We are not using the line text anymore, but only the length.
  5. 27 Nov, 2013 2 commits
  6. 29 Dec, 2012 1 commit
  7. 08 Dec, 2012 2 commits
  8. 06 Dec, 2012 1 commit
  9. 21 Oct, 2012 2 commits
  10. 23 Sep, 2012 1 commit
  11. 01 Apr, 2012 3 commits
  12. 11 Mar, 2012 1 commit
  13. 10 Mar, 2012 1 commit
  14. 03 Mar, 2012 2 commits
  15. 25 Feb, 2012 2 commits
  16. 07 Jan, 2012 1 commit
    • Thomas Freitag's avatar
      regtest: read stderr output before calling wait to fix a possible deadlock · c2378609
      Thomas Freitag authored
      In the way the scripts starts it subprocesses, stderr becomes buffered
      for the subprocesses. And unforunately, when the buffer limit is
      reached, the subprocess suspends it work until it can print again on
      stderr. That's why the python script runs into a deadlock when the
      subprocess produces a lot of error messages. A small rearrange of the
      commands, first read the stderr output and then wait that the subprocess
      ends, will remove this deadlock.
  17. 06 Nov, 2011 1 commit
  18. 01 Nov, 2011 1 commit
  19. 20 Sep, 2011 1 commit
  20. 19 Sep, 2011 2 commits
  21. 18 Sep, 2011 3 commits
  22. 13 Sep, 2011 1 commit