Question

I would like to implement a tool that generates graphs whose memory will be allocated on a data structure so called “Tape”. You can think of a Tape as an array of elements, each of which holds “Node ID”, links to its “Parent Node” as well as its “Child Nodes”.

What I am looking for is an approach in which identifying available slots in the array is cheap so that when a new node is to be added, an empty slot can be quickly identified.

And what if I implemented the Tape using a dynamic array? In the situation where the size of the array needs resizing, can I avoid copying the entire Tape over to a newly allocated array?

Anyone here has any idea ?

Answer 1

I assume that you want to allocate a big ‘Tape’ beforehand of e.g. thousands of nodes.

You should combine 2 concepts:

• First store the last used entry on your tape. The next time a new entry is needed, just pick the one just after the last used one.
• Second keep a free list. Use the first 4 bytes (or 8 bytes in 64-bit) of your tape entry as a pointer to the next free entry. The beginning of the tape should point to the first free entry.

Whenever a entry on the Tape is freed, add it to the free list.

Whenever a new entry is needed in the tape:

• check whether there are elements in the free list. if there are take the first entry and remove it from the free list
• if the free list is empty, use the last used entry and take the one immediately after it.

You can also combine this with a reallocation scheme, where you keep the total allocate size of your tape and reallocate if the last used entry reaches the end of the tape.