Permalink
Commits on Apr 2, 2011
  1. Add define for Rubinius, because we're a nicely behaving extension

    This can greatly improve performance in Rubinius. We don't change
    a string buffer from Ruby nilly willy.
    dbussink committed Apr 2, 2011
  2. Fix subsecond Time handling

    dbussink committed Apr 2, 2011
  3. Update copyright dates

    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Apr 2, 2011
Commits on Apr 1, 2011
  1. [do_jdbc] Fix Reader#fields inconsistency w/ column aliases

    * Previously C drivers were respecting aliases, whereas Java drivers
      were not, by virtue of our use of ResultSetMetaData#getColumnName
      rather than ResultSetMetaData#getColumnLabel.
    
    * This didn't become apparent because MySQL's Connector/J 5.0.x did
      not respect the JDBC Spec and followed aliases when
      ResultSetMetaData#getColumnLabel was invoked.
    
      This only became apparanet when we upgraded our versions of MySQL
      Connector/J to 5.1.x and found that dm-migrations broke. This
      necessitated a workaround – see dm-migrations commit # bd32f73:
    
      datamapper/dm-migrations@bd32f73
    
    * This commit should obviate need for this workaround.
    
    * This commit also adds explicit specs for MySQL's database metadata
      SQL queries, to avoid future regressions.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Apr 1, 2011
  2. Revert "[do_jdbc] Added dynamic proxy for various classes"

    * While proxing provides potential for instrumentation, for the
      time-being its not something we're using and adds another layer of
      complexity to the code-base (which if anything we should working on
      simplifying).
    
    This reverts commit 82644ec.
    
    Conflicts:
    
    	do_mysql/ext-java/src/main/java/do_mysql/MySqlDriverDefinition.java
    	do_sqlserver/ext-java/src/main/java/do_sqlserver/SqlServerDriverDefinition.java
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Apr 1, 2011
  3. Cleanup whitespace

    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Apr 1, 2011
  4. Update .gitignore for Eclipse project files

    * Begin restructuring .gitignore file.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Apr 1, 2011
Commits on Mar 29, 2011
  1. [all java] Mark JDBC URI specs as temp. pending

    * Our JDBC and JNDI URI handling is currently horribly broken, and
      awaiting a refactor of all URI handling code, we shoudln't even
      pretend to support these types of URIs.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Mar 29, 2011
  2. [do_postgres] Fix regression in Result / RETURNING specs

    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Mar 29, 2011
  3. [all] Move back to RSpec from Bacon

    * See previous commit # 09effb7 for explanation.
    
    * An initial revert of the port to Bacon, rebasing for changes since
      commit # 9d9d4cc, along with some minor cleanup and necessary changes
      for RSpec 2.0.
    * This commit doesn't attempt to exploit the optimal, most idiomatic
      RSpec 2.0 usage, but hopefully lays the groundwork for further
      improvement to specs.
    
    This (mostly) reverts commit 9d9d4cc.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Mar 29, 2011
Commits on Mar 25, 2011
  1. [data_objects] Move back to RSpec from Bacon

    * data_objects specs only (not shared specs).
    
    * Bacon is an excellent, compact and understandable testing framework
      which I hold in the highest regard. But the reasons for using Bacon
      no longer as valid as they were almost 18 months ago – namely that
      RSpec would not run on rbx at time:
    
      Bacon Pros:
      - faster (at least it used to be, haven't benchmarked against RSpec 2)
      - more hackable
      - will run on more alternative implementations (unsure if RSpec will
        now run on MacRuby, for example)
    
      RSpec Pros:
      - ubquity: we use it elsewhere in DataMapper
      - API niceties and more syntatical sugar
    
    This (mostly) reverts commit 355b302.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Mar 25, 2011
  2. Revert "workaround a bug in jruby in 1.9 mode"

    This reverts commit 711a63f.
    mkristian committed Mar 25, 2011
Commits on Mar 24, 2011
  1. fixed title

    mkristian committed Mar 24, 2011
  2. switch to latest jruby-1.6.0

    mkristian committed Mar 24, 2011
Commits on Mar 19, 2011
  1. [do_jdbc] Fix for JRuby 1.6.0RC2 in 1.9 mode

    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Feb 25, 2011
Commits on Mar 7, 2011
  1. [do_oracle] Fix uninitialized constant ConnectionError

    * Raise LoadError when JDBC drivers not found, as it seems more
      appropriate.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Mar 7, 2011
Commits on Feb 28, 2011
  1. Revert "[all multi] Downgrade rake-compiler to fix JRuby compile"

    This reverts commit 766d7a0.
    
    Conflicts:
    
    	do_mysql/Gemfile.lock
    	do_mysql/do_mysql.gemspec
    	do_oracle/do_oracle.gemspec
    	do_postgres/Gemfile.lock
    	do_postgres/do_postgres.gemspec
    	do_sqlite3/Gemfile.lock
    	do_sqlite3/do_sqlite3.gemspec
    myabc committed Feb 25, 2011
  2. Give failures to connect one last shot with UTF-8 encoding.

    Previously we were more selective about when we would attempt this; bad
    encodings were indicated through text inside the exception text.  However this
    seems to have changed lately, and failures to connect are surfaced without much
    more detail about the underlying why.
    
    So we just try once more with UTF-8 encoding (default) no matter what.  This is
    what the do_{adapter} specs suggest is the intended behaviour anyway.  If the
    connection failed for a reason other than encoding, then it'll fail again and
    the user will end up with the same error anyway.
    jpr5 committed Feb 28, 2011
Commits on Feb 27, 2011
Commits on Feb 25, 2011
  1. [all] Remove Gemfile.lock

    * Keeping Gemfile.lock in the repository is only best practice for
      applications and *not* libraries.
    * See also dm-core commit # 4386ff9.
    
    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Feb 25, 2011
  2. [do_oracle] Cleanup unused import

    Signed-off-by: Alex Coles <alex@alexbcoles.com>
    myabc committed Feb 21, 2011
Commits on Feb 21, 2011
  1. Fix more Connection Pool leaks in tests.

    Building on Dirkjan's d19180f, this solves a few more connection pool leaks by
    ensuring that all `before'/`after's that acquire/release from the pool are
    inside `describe' blocks (they shouldn't be immediately inside `shared' blocks).
    
    Currently triggering via fails on a Linux build box during the "parallel query"
    spec, which launches a 4 threads that use SQL to sleep 1, and expects them all
    to finish within 2 seconds.
    
    The few of us testing haven't seen this yet probably due to the subtle
    Dir[**/spec_*.rb] path-glob-order-relies-on-inode-order problem: depending on
    the order the spec files are run, the "parallel query" spec will either come
    earlier or later, and the earlier it is, the less likely it is to trigger (being
    the largest one-time consumer of the 8 maximum allowed pooled connections).
    jpr5 committed Feb 21, 2011
Commits on Feb 16, 2011