Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
branch: pls_rstar_hacks
Commits on Jul 28, 2010
  1. @cxreg
Commits on Jul 27, 2010
  1. @timbunce @jnthn

    added my project

    timbunce authored jnthn committed
  2. @moritz

    track changed auth name

    moritz authored
  3. @colomon

    Add Math::Polynomial.

    colomon authored
  4. @colomon
Commits on Jul 26, 2010
  1. @colomon

    Add Math::Vector.

    colomon authored
  2. @snarkyboojum
  3. @snarkyboojum
Commits on Jul 24, 2010
  1. @moritz
  2. @moritz

    switch to .IO file tests

    moritz authored
  3. @supernovus
Commits on Jul 14, 2010
  1. @jnthn

    Add some logic to at least identify if we have nmake rather than make…

    jnthn authored
    …, which makes things another notch more portable. Still some way off working.
  2. @jnthn

    Start to deal with some of the excessive UNIX-isms. This gets rid of …

    jnthn authored
    …the reliance on the find command, which means we can actually generate a Makefile now for projects without one. It's broken, but it's a step forward. Also replace one call out to mkdir with the Perl 6 command.
Commits on Jul 13, 2010
  1. @ingydotnet

    Add yaml-pm6 to list

    ingydotnet authored
Commits on Jul 10, 2010
  1. @moritz

    add svg to projects.list

    moritz authored
  2. Add bennu to project list.

    Tyler Leslie Curtis authored
Commits on Jul 9, 2010
  1. @moritz
Commits on Jul 6, 2010
Commits on Jul 4, 2010
  1. [proof-of-concept] announce 'Loading ecosystem'

    authored
    This is nice-ish, because it takes a while to do so. In practice, a load
    will only happen if a fetch is required. Since it loads a non-small JSON
    file, it takes a couple seconds. During this time, it now says what it's
    doing.
  2. [proof-of-concept] de-cluttered '[done]' output

    authored
    It's now '[ done ]', and the actual information is easier to read. Ditto
    '[ FAIL ]'.
  3. [proof-of-concept] Makefile gets a relative PERL6LIB

    authored
    This makes things easier. It feels oddly wrong, but I can't see why, and
    it actually works.
Commits on Jul 3, 2010
  1. [proof-of-concept] take skip-test into account

    authored
    In the diagnostics, the "which state broke" prediction now takes :$skip-test
    into account in the case when the status is left at "built".
  2. [proof-of-concept] fixed .pm file opening

    authored
    The ufo part of pls was under-tested, and blew up on first use. It should
    open a path not relative to the cached project, but relative to the pls
    dir. Also decided that a bare &open is too dangerous to have lying around,
    and surrounded it with a try {} block.
  3. [proof-of-concept] better error diagnostics

    authored
    The executable now explains possible action that can be taken upon error.
    It also leaves a more aptly named .log file after itself (on the pattern
    of 'fetch-json.log'), because I realized several steps can fail and leave
    several logfiles, which shouldn't be overwriting one another.
  4. [proof-of-concept] write to log instead of /dev/null

    authored
    When we run commands, write the commands to a logfile pls.log rather than
    to /dev/null. This has the advantage of working on Windows, which doesn't
    have a /dev/null. If the command succeeds, we remove the log file. The
    log file is always written to the pls directory.
  5. [proof-of-concept] s:4x/but/does/

    authored
    The difference between the C<but> and C<does> operators is that the latter
    makes a clone of the object before mixing in the role. If we're not interested
    in keeping the original, un-mixed-in object, we might as well use C<does>.
  6. [proof-of-concept] removed noisy output, added nice output

    authored
    Now each action takes one line, and gives nicely formatted output, totally
    shielding the end user from the harshness of the Real World.
  7. [proof-of-concept] suppressed all command output

    authored
    Ultimately we'll want to log things, but that's slightly lower prio right now.
  8. [proof-of-concept] lazily load ecosystem

    authored
    Since it takes a bit of time to load a non-small JSON file, we defer this
    task to when it's actually needed.
  9. [App::Pls] changed build strategy

    authored
    Instead of moving all dependencies of a project in lockstep with the project
    during build-test-install, Yapsi now sees to it that all dependencies
    are installed before building the dependent project. Suggested by jnthn++,
    and more in line with how CPAN does things. Makes more sense in many ways.
    
    A test started failing, but it's most likely due to the closures cloning bug
    in Rakudo. Skipped it for the time being.
Commits on Jul 2, 2010
  1. [.gitignore] Makefile

    authored
Something went wrong with that request. Please try again.