odbc 1.6.0

New features

• A hex logo has been added to the package (@edgararuiz, #824).

• databricks() and snowflake() detect viewer-based credentials on
  Posit Connect (@atheriel, #853 and #894).

• odbcEditDrivers(), odbcEditSystemDSN(), and odbcEditUserDSN()
  create shortcuts for file.edit(odbcListConfig()[[i]]) on macOS and
  Linux (@simonpcouch, #827).

• New redshift() helper simplifies connections to Amazon Redshift clusters,
  particularly with IAM credentials (@atheriel, #879).

Bug fixes and minor improvements

• Error messages rethrown from drivers no longer become garbled when raw messages
  contain curly brackets (@simonpcouch, #859).

• SQL Server drivers installed with homebrew are now discovered automatically
  on ARM macOS (@stevecondylios, #893).

• databricks(workspace) removes trailing slashes automatically, preventing
  uninformative driver messages (@simonpcouch, #827).

• dbConnect() displays clear errors when unixODBC isn't found on macOS and
  Linux (@simonpcouch, #782).

• dbConnect() adds name_encoding to complement encoding for separate
  column content and name encoding (#845).

• dbListTables() and dbExistsTable() improve temp table support with DB2 (#823).

• Netezza character and boolean columns receive improved data type
  inference (#847, #850).

• Oracle DATE and TIMESTAMP(n) targets handle writes correctly with
  batch_size > 1 (#810).

• snowflake() checks runtime driver configuration on macOS (#857).

• snowflake() accepts uid without pwd when using
  authenticator = "externalbrowser" or authenticator = "SNOWFLAKE_JWT"
  (@simonpcouch, #817 and #889).

• Resolved SQL Server data truncation in temp tables with FreeTDS (#866),
  writing issues with SIMBA drivers (#816), hms data roundtrip (#887),
  and precision when writing to DATETIME2 targets (#793).

odbc 1.5.0

Major changes

• New function snowflake() makes it easier to connect to Snowflake,
  automatically handling authentication correctly on platforms that provide
  Snowflake-native OAuth credentials (@atheriel, #662).

• Long running queries can now be interrupted using Ctrl-C. This
  feature is enabled by default in interactive sessions. It can be
  controlled by the interruptible argument to dbConnect() or by the
  global option odbc.interruptible. Should be considered experimental---if
  you experience problems please file an issue on the package's GitHub
  repository (#796).

Minor improvements and bug fixes

• Improved argument checking and transitioned to the cli package for
  formatting most existing error messages (@simonpcouch, #781, #784, #785, #788).

• Raises "Cancelling previous query" warnings from R rather than from Rcpp when
  a connection has a current result to avoid possible incorrect resource
  unwinds with options(warn = 2) (#797).

• Adjusted the default batch_rows value for dbWriteTable() and dbBind()
  methods. odbc 1.3.0 changed the default value from 1024 to NA, which sets the
  batch size to be the length of the input. While this addressed issues for
  some drivers when 1024 was greater than the number of rows, it also led
  to excessive memory consumption when the input was very large for some other
  drivers. The package will now interpet NA as the minimum of 1024 and the
  length of the input (@simonpcouch, #774).

• The encoding of non-ASCII column names of SQL results is now always converted
  to UTF-8. (@shrektan, #430)

• Improved error messages when the encoding of client and db-server are
  different. (@shrektan, #432)

• dbListFields() now works with Id() and SQL() identifiers (#771).

• Transitioned odbcDataType() to use S4 for consistency. S3 methods defined
  locally will need to be rewritten (@simonpcouch, #701).

• The "OdbcConnection" method for dbQuoteIdentifier() will no longer
  pass x to encodeString() before returning, for consistency with the
  default implementation in DBI (@simonpcouch, #765).

• A bug in the implementation of a new feature introduced in 1.4.2, where the
  package would automatically set the ODBCSYSINI environmental variable when
  using the unixODBC driver manager, was fixed; that environmental variable
  will now actually be set on package load (@simonpcouch, #792).

Driver specific changes

• databricks() will now automatically configure the needed driver and driver
  manager on macOS (@simonpcouch, #651).

• databricks() now picks up on Posit Workbench-managed Databricks
  credentials when rendering Quarto and RMarkdown documents in RStudio
  (@atheriel, #805).

• Improved performance on write with Snowflake (#760).

• Resolved issue when previewing tables using the RStudio Connections pane with
  Teradata (@simonpcouch, #755).

odbc 1.4.2

• dbAppendTable() Improve performance by checking existence once (#691).

• dbConnect() no longer automatically escapes suspicious characters
  (since there doesn't seem to be a consistent way to do this across drivers)
  but instead points you to quote_value() which applies a heuristic
  that should work for most drivers (#718).

• New wrapper for dbExecute() that sets immediate = TRUE if you are
  not supplying params. That should yield a small speed boost in
  many cases (#706).

• dbSendQuery() once again defaults to immediate = FALSE (since if you're
  using it instead of dbGetQuery() you're likely to be using it with
  dbBind()). (#726).

• Deprecated odbcConnectionColumns() (in favor of dbListFields()),
  odbcConnectionActions(), and odbcConnectionIcon() (@simonpcouch, #699).

• Backend specific changes:

  • databricks: Fix schema enumeration in connections pane
    (@detule, #715).

  • Oracle: use more reliable technique to determine user/schema name (#738),
    and fix dbExistsTable() when identifier components contain _
    (@detule, #712).

  • SQL Server: improvements to dbExists() (@meztez, #724) and
    dbListTables() (@simonpcouch, #509) for temporary tables.
    It now uses column type "BIGINT" integer64 objects.

  • SQL Server with freetds driver: no longer crashes when executing multiple
    queries (@detule, #731).

  • Teradata: Fix usage of exact argument in internal methods (@detule, 717).

• On MacOS and Linux, the package will now automatically set the ODBCSYSINI
  environmental variable when using the unixODBC driver manager. ODBCSYSINI
  will not be changed if it exists already (@simonpcouch, #709).

odbc 1.4.1

• New odbcListConfig() lists configuration files on Mac and Linux
  (@simonpcouch, #565).

• databricks() now works with manually supplied pwd and uid (#690).

• Oracle: uses correct parent class (#685).

• SQL Server: correctly enumerate schemas across databases in connections pane
  (@detule, #527).

• SQL Server: now uses column type "BIGINT" integer64 objects
  (@simonpcouch, #698).

odbc 1.4.0

Major changes

• New odbc::databricks() makes it easier to connect to Databricks,
  automatically handling many common authentication scenarios (@atheriel, #615).

• dbListTables(), dbListFields() and dbExistsTable() automatically
  escape underscores in identifier arguments. This leads to substantial
  performance improvements for some backends (e.g. snowflake)
  (@detule, @fh-afrachioni, #618).

• dbGetQuery() and dbSendQuery() now set immediate = TRUE if you are
  not using a parameterised query. That should yield a small speed boost in
  many cases (#633).

Minor improvements and bug fixes

• Increased the minimum required R version from 3.2.0 to 3.6.0
  (@simonpcouch, #629).

• S4 classes for the most database drivers are now exported, make it possible
  to use in other packages (#558).

• ODBC errors are now spread across multiple lines, making them easier to
  read (@detule, #564).

• DBI::dbConnect(odbc::odbc()) now gives a clear error if you supply multiple
  arguments with the same name when case is ignored (#641).

• DBI::dbConnect(odbc::odbc()) now automatically quotes argument values that need
  it (#616).

Driver specific changes

• Oracle: Fix regression when falling back to odbcConnectionColumns() to
  describe column data types (@detule, #587)

• Spark SQL: Correctly enumerate schemas away from the current catalog
  (@detule, #614)

• Snowflake: improved translation from R to snowflake types (@meztez, #599).

• SQL Server

  • Improved handling for local temp tables in dbWrite(), dbAppendTable(),
    and dbExistTable() (@detule, #600)

  • Specialize syntax in sqlCreateTable to avoid failures when
    writing to (new) local temp tables. (@detule, #601)

• Teradata: Improved handling for temp tables (@detule and @But2ene, #589, 590)

odbc 1.3.5

• Various fixes for R CMD check.

• The argument order for sqlCreateTable() methods avoiding weird argument
  mismatch problems (#448).

• Fixed truncation when retrieving unicode data stored in
  VAR/CHAR columns (@detule, #553).

• Fixed issue related to fetching zero rows (@detule, #528).

• Backend specific improvements:

  • MYSQL: Fixed retrieving results from stored procedures (@detule, #435).
  • Oracle: performance enhancements (#577)
  • Spark: Changed data type for date-times from DATE to TIMESTAMP
    (@schuemie, #555).

odbc 1.3.4

• Optimized table preview methods. Enhances large table preview
  within RStudio IDE (@detule, #525).

• Added attributes parameter to dbConnect(...). Can be used to pass a
  token and authenticate against Azure AD when using Microsoft SQL Server.
  Documented in ?ConnectionAttributes (@detule, #521).

• Fix length(x) = 3 > 1' in coercion to 'logical(1) warning in connection
  observer (@meztez, #494).

odbc 1.3.3

• Hadley Wickham is now the maintainer.

odbc 1.3.2

• New odbcDataType.Snowflake() method for Snowflake databases. (@edgararuiz, #451)

odbc 1.3.1

• Fixed warnings about anonymous unions (@detule, #440)
• Fixed invalid descriptor issues when retrieving results from SQL Server +
  Microsoft's ODBC driver, using parametrized queries. (@detule, #414)
• Fixed null handling in SQL Server / Azure result sets retrieved with
  Microsoft's ODBC driver. (@detule, #408)
• Hive uses C-style escaping for string literals (single quotes are
  backslash-escaped, note single quote-escaped). dbQuoteString now respects
  this when called on a connection of class Hive. (@rnorberg, #184)
• When calling sqlCreateTable(con, ..., temporary = TRUE) and con is a
  connection of class DB2/AIX64, the CREATE TABLE statement that is generated
  properly creates a temporary table in DB2. The statement begins with
  DECLARE GLOBAL TEMPORARY TABLE
  and ends with ON COMMIT PRESERVE ROWS (DB2's default behavior is
  ON COMMIT DELETE ROWS, which results in the inserted data being
  deleted as soon as dbWriteTable completes). (@rnorberg, #426)