Things may still change

Very untested as of now

c-vector

Compile Time Generic Dynamic Vector in C.

First Things First

On it's own, no code is being compiled. You first have to include and implement a vector of your desires. For that, there are two macros.

#include "vec.h"
VEC_INCLUDE(N, A, T, M);
VEC_IMPLEMENT(N, A, T, M, F);

1. N - Name - the resulting name of the vector struct
2. A - Abbreviation - functions get prefixed with that
3. T - Type - the type of your elements within the vector
4. M - Mode - storage type, either BY_VAL (by value) or BY_REF (by reference)
5. F - Free - provide a freeing function for your elements, if available

Example

Basic Vectors

VEC_INCLUDE(VecU8, vec_u8, unsigned char, BY_VAL);
VEC_IMPLEMENT(VecU8, vec_u8, unsigned char, BY_VAL, 0);

VEC_INCLUDE(VecSize, vec_size, size_t, BY_VAL);
VEC_IMPLEMENT(VecSize, vec_size, size_t, BY_VAL, 0);

Struct Vectors

VEC_INCLUDE(Vec2U8, vec2_u8, VecU8, BY_VAL);
VEC_IMPLEMENT(Vec2U8, vec2_u8, VecU8, BY_VAL, vec_u8_free);

VEC_INCLUDE(Vec2U8, vec2_u8, VecU8, BY_REF);
VEC_IMPLEMENT(Vec2U8, vec2_u8, VecU8, BY_REF, vec_u8_free);

Design Considerations

• Both BY_VAL and BY_REF can be used on either basic types or (complex) structs. Either way, if you (e.g.) push back a value, its content is copied
• Generally speaking, I'd use the latter when dealing with structs where sizeof struct > sizeof(void *).
• I went ahead and compared the very basic performance between the two approaches here.

Metaprogramming

Advantages

• Compiler optimized code
• Flexibility, reusability (generics / templates)
• Type safety

Disadvantages

• Maintenance challenges
• Debugging difficulties
• Compilation overhead

How to use it

Examples

• $ cd examples && make (binaries in subfolder "bin") -> WIP, I want to add more examples

Tests

• $ cd test && make (binaries in subfolder "bin") -> WIP, I want to add more tests to make the vector bug free
• Ignore the compile_flags.txt files -> I added those only so that my LSP knows what's up.

Available Functions

The A## means the A specified in the two macros.

• A##_zero set vector to zero without freeing
• A##_clear clear vector but keep capacity
• A##_length return length of of the vector in items
• A##_capacity return capacity of the vector
• A##_empty return if it's empty
• A##_resize resize vector to optimal given length in items
• A##_free free all memory used
• A##_reserved return bytes allocated
• A##_reserve reserve more memory in items
• A##_shrink shrinks vector to minimal possible capacity
• A##_set_at overwrite item at index, free any previous item
• A##_push_front insert item in the front
• A##_push_back insert item at the back
• A##_push_at insert item after index
• A##_pop_front pop item in the front
• A##_pop_back pop item at the back
• A##_get_at get item at index
• A##_get_front get item in the front
• A##_get_back get item at the back
• A##_swap swap two items by index
• A##_reverse reverse the vector
• A##_iter_begin return beginning iterator
• A##_iter_end return end of iterator

Additional Settings

There are various settings one can adjust to fit the vector to their needs. To use those, I strongly recommend the following:

• In header files (using the INCLUDE macro) -> #define the respective settings before including vec.h. Before the end of your header #undef them, so that other possible vectors that may create a vector of such a "modified" vector with custom settings reverts back to the defaults
• In source files (using the IMPLEMENT macro) -> #define the respective settings after you included your custom vector header and have them be equally set.

Settings list

• VEC_SETTINGS_KEEP_ZERO_END number; specify how much zero memory should be kept at the end when reserving memory (e.g. string implementation)
• VEC_SETTINGS_STRUCT_ITEMS literal; specify the name of the items placeholder to something else, if you so desire (e.g. string implementation, where it makes more sense to use another literal besides the previously mentioned for the string placeholder)
• VEC_SETTINGS_DEFAULT_SIZE number; specify how many item spaces shall be reserved minimally

todo so I don't forget

• add pop_at pop item at index
• add emplace insert item before index
• add stuff that allows to pop/push/insert/emplace an entire array, or a subsection...
• if we pushed from the front, then N.first can actually be reduced, instead of moving everything one back!
• bundle the snippets I copied around used for freeing into it's own function
• comparing stuff requiring a comparing function: cmp, find, match, rfind, rmatch, invert
• cat, back/at/front, pop_at, pop_slice, apply