|
1 | 1 | # odbc (development version) |
2 | 2 |
|
3 | | -* odbc can now be compiled again with Rtools35 (gcc 4.9.3) (#383) |
4 | | -* `invalid descriptor` issues due to out of order retrieval of long columns are now avoided by unbinding any nanodbc buffer past the long column. Performance for the unbound columns in these cases will be reduced, but the retrieval will work (@detule, #381) |
5 | | -* `dbWriteTable()` now handles `data.table::IDate()` objects (#388) |
6 | | -* `odbcConnectionColumns()` now works as intended with `DBI::Id()` objects (#389) |
7 | | -* Oracle connections now support `Date` and `POSIXct` types via `DATE` and `TIMESTAMP` data types (#324, #349, #350) |
8 | | -* Oracle connections now use VARCHAR2 rather than VARCHAR, as recommended by Oracle's documentation (#189) |
9 | | -* `dbWriteTable(field.types=)` now issues a warning rather than an error for missing columns (#342) |
| 3 | +## Major changes |
| 4 | +* odbc can now be compiled again with Rtools35 (gcc 4.9.3) on Windows (#383) |
| 5 | +* `invalid descriptor` errors from drivers such as Microsoft SQLServer driver and the freeTDS driver which do not support out of order retrieval are now avoided. |
| 6 | + This is done by unbinding any nanodbc buffer past the long column. |
| 7 | + Performance for the unbound columns in these cases will be reduced, but the retrieval will work without error (@detule, #381) |
| 8 | +* `dbBind()` and `dbFetch()` now support multiple result sets (@vkapartzianis, #234) |
| 9 | + |
| 10 | +## Minor improvements and fixes |
10 | 11 | * New `dbAppendTable()` method for OdbcConnection objects (#335) |
| 12 | +* `dbQuoteIdentifier()` now uses the input names (if any). |
11 | 13 | * `dbWriteTable()` and `dbBind()` now default to a `batch_rows` of `NA`, which sets the batch size to be the length of the input. |
12 | 14 | This avoids problems with drivers that don't support batch sizes larger than the input size. |
13 | 15 | To restore the behavior prior to this release pass `batch_rows = 1024` or set `options(odbc.batch_rows = 1024)` (#391). |
14 | | -* `dbBind()` and `dbFetch()` now support multiple result sets (@vkapartzianis, #234) |
| 16 | +* `dbWriteTable()` now handles `data.table::IDate()` objects (#388) |
| 17 | +* `dbWriteTable(field.types=)` now issues a warning rather than an error for missing columns (#342) |
| 18 | +* `odbcConnectionColumns()` now works as intended with `DBI::Id()` objects (#389) |
15 | 19 | * `dbFetch()` now verifies that `n` is a valid input. |
16 | | -* `dbQuoteIdentifier()` now uses the input names (if any). |
17 | 20 | * Dates are now always interpreted as being in the database's local time zone, regardless of the `timezone` parameter (#398) |
| 21 | +* Oracle connections now support `Date` and `POSIXct` types via `DATE` and `TIMESTAMP` data types (#324, #349, #350) |
| 22 | +* Oracle connections now use VARCHAR2 rather than VARCHAR, as recommended by Oracle's documentation (#189) |
18 | 23 |
|
19 | 24 | # odbc 1.2.3 |
20 | 25 |
|
|
0 commit comments