CMake support and some improvements

What's new?

• CMake support
• jules::drop
• jules::index_view

Fixes

Some small fixes

(Almost)-complete Array module.

🎉 Celebrating 4K lines of code!

What's new?

In this release, all the basic and fundamental functionalities of the Array module have been implemented.
They are not still documented, but I do not intend to implement major changes for at least 5 years.

Many small utilities have been implemented and the test coverage has increased a lot.

Changes

The major change in this release is that the operator() of arrays has been removed.
Instead, an operator[] is used for each dimension. Such change brings not only a more familiar
syntax, but also many performance improvements.

New utilities for arrays and Range TS changes.

In this patch, some math and numeric utilities for arrays were added and some changes
were made to comply with the newest Range TS.

Bug fixes

• Complying with the newest Range TS.
• Fixed normal distribution p.d.f. formula.

New features

• Added sqrt, log, var, sd, mean, freq, which.

Coercion improvements and more stable Array module

Bug fixes

• Jules now compiles with -pedantic.
• When using BLAS-optimized operation, the maximum number of rows and columns is checked.
• Fixed virtual destructor for column_model.
• Fixed some narrowing integer conversions in internal code.
• Fixed critical bug while checking slicing bounds.
• Removed useless slice::iterator::operator->.
• Preventing view of temporary columns.
• Fixed several bugs in empty columns.
• More documentation.

New "stable" features/changes

• Added operations for contiguous_array.
• Better defer utilities (using macros, though).
• Added uninitialized constructor for base_array.
• base_array is value-initialized by default instead of default-initialized.

New "experimental" features/changes

• Calling column::data does not coerce the column type.
• Added to_matrix for dataframes.
• to_view now performs automatic coercion in columns.
• Added index_t, int and unsigned coercion rules.
• Learned coercions must be explicitly defined by the user.
• Improved column_model to support partial cloning and resizing.
• Added initial dataframe action support, with head implementation.

Initial Dataframe support

New features

• Initial dataframe and column support.
• Added jules::meta utilities.
  • It reduces range-v3 dependencies.
• Added jules::contiguous_array for non-owning contiguous storage.
  • Used in column views.

Improved features

• jules::debug supports throwing handlers.
  • Useful for code coverage and checks in the Dataframe module.
• Bug fixes on Numeric module.
• New utilities in the Numeric module (incomplete, but working great).

Initial Random support

This version includes some fixes on the Array module and the initial support for Base/Random module.

v0.1

Special thanks to @jimmyskull for hunting bugs.

Changes since v0.1-rc:

• Deprecated slice::dimensions;
• as_vector for arrays of higher order;
• Iterators do not inherit std::iterator anymore;
• array::fill for fast filling;
• Initial BLAS support;
• Faster indirect slicing.

v0.1-rc

• Only the array module is working at the moment (along with fundamental functionalities of modules base and core).
• The files under directories other than array, base and core shouldn't be included.
• No new features will be incorporated into array classes before v0.2 series.
• Features in array, base, core are guaranteed not to be removed before v0.3 series.
  • If necessary, they will be marked [[deprecated]] in v0.2 series.
• Algorithms and utilities included in base/numeric and array/numeric are not considered documented features yet, and thus, they can change during v0.1 series.
  • as_vector, to_vector, and seq are exceptions to this rule and will not change in the v0.1 series.