Skip to content
Commits on May 31, 2016
  1. implement a basic generic table viewer

    Introduce a new function, renderTableViewer(), which renders a simple
    HTML table from the given array of data rows and the declaration of
    the table columns. Update renderCactiConfig() and renderMuninConfig()
    to look at it in action.
    committed May 31, 2016
  2. implement the zebra TR pattern with CSS, pt. 4

    * pi.css: add a "zebra0" CSS class for tables w/o the TH row at top
    * renderObjectLogEditor(): switch to the new class
    committed May 31, 2016
  3. remove an unused global declaration

    committed May 31, 2016
  4. implement the zebra TR pattern with CSS, pt. 3

    * renderAttributes()
    * allObjectLogs()
    * renderVirtualResourcesSummary()
    committed May 31, 2016
Commits on May 27, 2016
  1. @andriyanov

    Merge pull request #161 from xornet-sl/maintenance-0.20.x

    snmp.php: use showError instead of showFuncMessage
    andriyanov committed May 27, 2016
  2. @xornet-sl

    remove unused message codes

    xornet-sl committed May 27, 2016
  3. Travis CI: do load the sample data, but later

    Loading the sample data dump into a freshly generated RackTables
    database is actually a useful test by itself so it wasn't right to drop
    it in the previous commit. As the data can interfere with the tests,
    these just should be run before loading the dump, not instead
    (especially as the unit tests are assumed to tear all their fixtures
    down as they complete). And as soon as the sample data dump cannot be
    easily unloaded, it does not belong to the express test set, which is
    purposed for repeated runs in the same working copy by a developer.
    In other words, it is a valid test that belongs only to a specific step
    of the Travis CI script.
    committed May 27, 2016
  4. @xornet-sl

    snmp.php: use showError instead of showFuncMessage

    now snmp.php can be included in script mode, not only in ophandler
    xornet-sl committed May 27, 2016
Commits on May 26, 2016
  1. Travis CI: don't load the sample data

    The sample data may interfere with the outcome of the tests in either
    way (false negative or false positive) and it would not reproduce
    consistently between the Travis CI temporary copy and the developer's
    working copy because of non-test data being different.
    To make this more deterministic it seems better to initialize the
    Travis CI database without the sample data and to expect all normal
    dependencies of a test to be managed by the test class methods. This way
    it should be easier to correlate the cause and the effect.
    committed May 26, 2016
Commits on May 25, 2016
  1. Travis CI: use SQL queries from the installer

    The SQL dumps in tests/data for each respective release contain the
    initial schema and minimal data plus the contents of
    init-sample-racks.sql on top. This way using even the most recent dump
    from that directory would (and in fact did) fail the builds made after
    changes to install.php. For example, a pull request that makes changes
    to the schema would likely fail the test as the database would belong to
    a release version.
    This commit changes the test setup script to initialize the database
    the same way as the installer would do, and to load the current
    init-sample-racks.sql after that. This allows to test a part of the
    actual path the installer takes and resolves the failed pull requests
    committed May 25, 2016
  2. dictionary: remove duplicate UCS servers

    This fixes a failure to install as the unique index was violated. For
    the time being also delete respective values from $ucsproductmap.
    committed May 25, 2016
Commits on May 23, 2016
  1. commitAddPort(): do not specify a default value

    In this case it served no valid purpose and just caused confusion.
    committed May 23, 2016
Commits on May 20, 2016
  1. tests: remove duplicate assertions

    Do not repeat in RackspaceFunctions::testRSU() what is already done in
    committed May 20, 2016
  2. use PDOStatement::fetchColumn() in more functions

    See commit aa06682.
    * getRowMountsCount()
    * getRackMountsCount()
    committed May 20, 2016
  3. tests: cover some rackspace functions

    * getRSUForRack()
    * getRSUForRow()
    * getRackMountsCount()
    * getRowMountsCount()
    committed May 20, 2016
Commits on May 19, 2016
  1. @bpothier
  2. stringForOption(): increase $maxlen (#1611)

    This is a fixup to commit dea303a.
    committed May 19, 2016
  3. fix location filter in the rackspace page (#1595)

    This fixes a bug introduced in commit d9bad31.
    * renderRackspace()
    committed May 19, 2016
  4. @github138
Commits on May 18, 2016
  1. fixup indentation

    committed May 18, 2016
  2. rectify renderIPAddressLog()

    Align columns to the left and dismiss the old row style management.
    committed May 18, 2016
  3. @github138
Commits on May 15, 2016
  1. @andriyanov

    Merge pull request #157 from etcet/maintenance-0.20.x

    Add 'live ports' support for fdry5 as well as support Brocade FCX 648 and FLS 648
    andriyanov committed May 15, 2016
Commits on May 12, 2016
  1. tests: cover more port-related functions

    * commitUpdatePort()
    * commitUpdatePortOIF()
    * linkPorts()
    * commitUpdatePortLink()
    * commitUnlinkPort()
    committed May 12, 2016
  2. refine commit 5504a99

    committed May 12, 2016
Commits on May 11, 2016
  1. conform to MySQL NO_ZERO_DATE SQL mode, pt. 1

    MySQL versions prior to 5.7 typically have NO_ZERO_DATE SQL mode
    disabled, whilst version 5.7 typically has it enabled. RackTables
    versions up to and including 0.20.11 contain a few SQL columns that have
    the default value set to "0000-00-00 00:00:00", hence those versions
    will fail to install on a MySQL server with NO_ZERO_DATE enabled.
    Interestingly, the same RackTables versions will continue to work if
    installed on a MySQL server without NO_ZERO_DATE and later changed to
    with it (likely as an implicit effect of MySQL upgrade from pre-5.7 to
    One possible solution to this problem could be to start managing the
    server SQL mode during DB connection setup. But as it is difficult to
    predict which way MySQL will evolve in this regard, a better long-term
    solution seems to be just not to use the zero date default value (and
    any other mode-dependent language features in general) at all.
    This commit changes the first of the two involved tables -- LDAPCache,
    in which the column last_retry is used only as input to TIMESTAMPDIFF(),
    which returns NULL when the argument is NULL or the zero date. This way
    the change should not be visible in the associated PHP code (though I
    have not tested it).
    * install.php
    * upgrade.php
    committed May 11, 2016
Commits on May 8, 2016
Something went wrong with that request. Please try again.