fetched editor changelog for 0.35

Author: gdementen

doc/source/changes/editor/version_0_35.rst.inc

@@ -0,0 +1,148 @@
+.. py:currentmodule:: larray_editor
+
+New features
+^^^^^^^^^^^^
+
+* allow displaying *many* more different kinds of objects, and not only arrays
+  from larray. One specific goal when developing this new feature was speed.
+  Most of these viewers should be fast (when at all possible), even on (very)
+  large datasets. We only support displaying (not editing) all the new types.
+
+  The following types are supported so far (but adding more is relatively easy):
+
+  * Python builtin objects:
+    - tuple (including named tuple), list (sequences), dict (mappings),
+      dict views, memoryview and array
+    - text and binary files
+  * Python stdlib objects:
+    - pathlib.Path
+      * if the path points to a directory, it will display the content of the
+        directory
+      * if the path points to a file, it will try to display it, if we
+        implemented support for that file type (see below for the list
+        of supported types).
+    - sqlite3.Connection (and their tables)
+    - pstats.Stats (results of Python's profiler)
+    - zipfile.ZipFile and zipfile.Path
+  * new objects from LArray: Axis, Excel Workbook (what you get from
+    larray.open_excel()), Sheets and Range
+  * IODE "collections" objects: Comments, Equations, Identities, Lists, Tables,
+      Scalars and Variables, as well as Table objects
+  * Pandas: DataFrame, Series and DataFrameGroupBy
+  * Polars: DataFrame and LazyFrame
+  * Numpy: ndarray
+  * PyArrow: Array, Table, RecordBatchFileReader (reader object for feather
+    files) and ParquetFile
+  * Narwhals: DataFrame and LazyFrame
+  * PyTables: File, Group (with special support for Pandas DataFrames written
+    in HDF files), Array and Table
+  * IBIS: Table
+  * DuckDB: DuckDBPyConnection and DuckDBPyRelation (what you receive from any
+    query)
+
+  File types (extensions) currently supported:
+    - Iode files: .ac, .ae, .ai, .al, .as, .at, .av, .cmt, .eqs, .idt, .lst,
+                  .scl, .tbl, .var
+    - Text files: .bat, .c, .cfg, .cpp, .h, .htm, .html, .ini, .log, .md,
+                  .py, .pyx, .pxd, .rep, .rst, .sh, .sql, .toml, .txt, .wsgi,
+                  .yaml, .yml
+    - HDF5 files: .h5, .hdf
+    - Parquet files: .parquet
+    - Stata files: .dta
+    - Feather files: .feather
+    - SAS files: .sas7bdat
+      It is limited to the first few thousand rows (the exact number depends on
+      the number of columns), because reading later rows get increasingly slow,
+      to the point of being unusable.
+    - CSV files: .csv
+    - Gzipped CSV files: .csv.gz
+    - Excel files: .xls, .xlsx
+    - Zip files: .zip
+    - DuckDB files: .ddb, .duckdb
+
+* the editor now features a new "File Explorer" (accessible from the "File"
+  menu) so that one can more easily make use of all the above file viewers.
+
+* added a new SQL Console (next to the iPython console) for querying Polars
+  structures (DataFrame, LazyFrame and Series) as SQL tables. The console
+  features auto-completion for SQL keywords, table names and column names
+  and stores the last 1000 queries (even across sessions). Recalling a query
+  from history is done with the up and down arrows and like in the iPython
+  console, it searches through history with the current command as prefix.
+  This console will only be present if the polars module is installed.
+
+* allow sorting some objects by column by pressing on an horizontal label.
+  This is currently implemented for the following objects:
+    - python built-in sequences (e.g. tuples and lists)
+    - python pathlib.Path objects representing directories
+    - LArray (only for 2D arrays)
+    - Pandas DataFrame
+    - Polars DataFrame and LazyFrame
+    - Narwhals LazyFrame
+    - SQLite tables
+    - DuckDB relations
+
+* allow filtering some objects by pressing on an horizontal label.
+  This is currently implemented for the following objects:
+    - Pandas DataFrame
+    - Polars DataFrame and LazyFrame
+    - DuckDB relations
+
+* allow comparing arrays/sessions with different axes in :py:obj:`compare()`.
+  The function gained ``align`` and ``fill_value`` arguments and the interface
+  has a new combobox to change the alignment method during the comparison:
+      - outer: will use a label if it is in any array (ordered like the first array).
+               This is the default as it results in no information loss.
+      - inner: will use a label if it is in all arrays (ordered like the first array).
+      - left: will use the first array axis labels.
+      - right: will use the last array axis labels.
+      - exact: raise an error when axes are not equal.
+  Closes :editor_issue:`214` and :editor_issue:`251`.
+
+* double-clicking on a name in the variable list will open it in a new window
+  (closes :editor_issue:`143`).
+
+
+Miscellaneous improvements
+^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+* made the editor interruptible by an outside program (i.e. made PyCharm stop &
+  restart buttons work directly instead of only when the editor receives the
+  focus again). Closes :editor_issue:`257`.
+
+* resize axes and vertical label columns automatically
+
+* string values are left aligned instead of right aligned
+
+* when comparing sessions via :py:obj:`compare()`, the color of arrays in the
+  list is now updated depending on the tolerance. To reflect that the tolerance
+  widget moved to the top of the interface. Closes :editor_issue:`201`.
+
+* :py:obj:`compare()` max difference is colored red when the difference is not 0
+
+* typing the name of a variable holding a matplotlib figure (or axes) in the
+  console shows it (previously, only expressions were displayed and *not*
+  simple variables). For example: ::
+
+     >>> arr.plot()
+
+  already displayed a plot, but: ::
+
+     >>> ax = arr.plot()
+     >>> ax
+
+  did not.
+
+* display matplotlib Figure objects in addition to Axes objects
+
+
+Fixes
+^^^^^
+
+* fixed :py:obj:`compare()` colors when the only difference is nans on either
+  side.
+
+* fixed :py:obj:`compare()` colors and max difference label when the only
+  differences are for rows where the value is 0 in the first array.
+
+* fixed single column plot in viewer when ticks are not strings