Permalink
Commits on Feb 26, 2015
  1. reset color on exit

    committed Feb 26, 2015
Commits on Sep 27, 2014
  1. (minor) fixup to tshrec

    committed Sep 27, 2014
Commits on Sep 14, 2014
  1. tshrec: record tsh sessions

    committed Sep 14, 2014
Commits on Sep 6, 2014
  1. make it easy to find Tsh.pm

    committed Sep 6, 2014
Commits on Apr 9, 2014
  1. @milki

    Test for cd -

    (minor fixup by sitaram)
    milki committed with Apr 3, 2014
  2. @milki

    cd recognizes - argument

    'cd -' returns to the previous directory.
    
    This is not valid if no directory change has happened before.
    milki committed with Apr 3, 2014
Commits on Mar 18, 2014
  1. tsh: added license info

    committed Mar 18, 2014
Commits on Mar 2, 2014
Commits on Dec 21, 2012
  1. remove obsolete note re gitolite test suite...

    thanks to Ralf H for catching this.  The v3 test suite is now "all tsh",
    so this comment was misleading.
    committed Dec 21, 2012
Commits on Dec 20, 2012
Commits on Nov 22, 2012
Commits on Nov 19, 2012
  1. make "prove" work

    committed Nov 19, 2012
Commits on Mar 15, 2012
  1. new cmp() function

    committed Mar 15, 2012
Commits on Jan 27, 2012
  1. use fixed timestamps (for predictable SHAs) only if running under har…

    …ness
    
    I'd like to be able to use tsh under normal conditions also, like just
    saying 'tsh empty' to get an empty commit, and in that case I'd like the
    commit to have the current time.
    
    When not under a harness:
    
      - $tick does not get defined, nor do GIT_COMMITTER_DATE and
        GIT_AUTHOR_DATE.
    
      - 'tc' (test-commit) uses the current time instead of our own internal
        epoch of 1310000000.
    
      - 'tt' (test-tick) will 'sleep 1' to make the actual time go forward.
    
    Note that harness commits move forward by 1 minute on a 'tt', but if we
    change that now it'll affect a lot of SHAs in a lot of test scripts.
    committed Jan 27, 2012
Commits on Jan 19, 2012
  1. make finding location of errors easier

      - when verbose=1, print $text also on fail
      - TSH_ERREXIT to emulate 'set -e' in bash; also prints fail info like
        TSH_VERBOSE=1
    committed Jan 19, 2012
Commits on Jan 17, 2012
  1. (minor docfix)

    committed Jan 17, 2012
  2. new run()

    committed Jan 17, 2012
Commits on Jan 16, 2012
Commits on Dec 14, 2011
  1. allow hashhash comments to show up on STDOUT

    this helps in all cases
    
      - progress for non-harness runs
      - silent for 'prove'
      - verbose for 'prove -v'
    committed Dec 14, 2011
Commits on Dec 12, 2011
Commits on Dec 11, 2011
  1. (minor)

    committed Dec 10, 2011
  2. added lines() function

    committed Dec 10, 2011
Commits on Dec 9, 2011
  1. stop using parse_line from Text::ParseWords...

    I reluctantly had to go back to a simplistic 'split()' with a negative
    look-behind to handle the simplest cases of an escaped ';'.
    
    The module would have been nice, but can't do what I want.  When the
    input is, say:
    
        abc; /\.foo/\; echo $HOME; ghi
    
    I want it to be parsed as the following commands:
    
        abc
        /\.foo/; echo $HOME
        ghi
    
    Notice I want the "\" retained in the regex, but not the one escaping
    the semicolon.
    
    But parse_line will either lose both the backslashes or retain both of
    them; you can't tell it to lose them only on the delimiter used.  Pity.
    
    You could argue that the correct solution is to let it lose all the
    delimiters and make the user add more.  This is certainly what *every*
    tool does, but it increases clutter in what I want to be a really really
    small language -- saying /\\\\.foo/ just doesn't have the same charm as
    /\.foo/ :)
    committed Dec 9, 2011
Commits on Dec 8, 2011
  1. (minor) buglet

    in a function that we have not yet chosen to document anyway :)
    committed Dec 7, 2011