ArrayDeque added

[HarsheetKakar]

References to other Issues or PRs or Relevant literature

Fixes #120

Brief description of what is fixed or changed

ArrayDeque added
small change in DODA instead of obj[0] it is now obj._data[0] in __new__

Other comments

[codecov]

Codecov Report

Merging #247 into master will increase coverage by 0.056%.
The diff coverage is 100.000%.

@@              Coverage Diff              @@
##            master      #247       +/-   ##
=============================================
+ Coverage   98.575%   98.632%   +0.056%     
=============================================
  Files           24        24               
  Lines         2106      2193       +87     
=============================================
+ Hits          2076      2163       +87     
  Misses          30        30               

Impacted Files	Coverage Δ
pydatastructs/linear_data_structures/arrays.py	100.000% <100.000%> (ø)
...datastructs/miscellaneous_data_structures/queue.py	100.000% <100.000%> (ø)

[czgdp1807]

Please follow the pattern currently there in queue.py. Can we avoid adding new classes?

[HarsheetKakar]

Please follow the pattern currently there in queue.py. Can we avoid adding new classes?

add it in the implementation? but the thing is deque is best of stack and queue and is implemented like DODA so I thought it doesnt fit in as another implementation of queue. I can change it if you want.

[czgdp1807]

but the thing is deque is best of stack and queue and is implemented like DODA so I thought it doesnt fit in as another implementation of queue. I can change it if you want.

It should be changed. A doubly ended queue is anyway a queue. You can re-define some methods in the implementation to make things work.

[HarsheetKakar]

but the thing is deque is best of stack and queue and is implemented like DODA so I thought it doesnt fit in as another implementation of queue. I can change it if you want.

It should be changed. A doubly ended queue is anyway a queue. You can re-define some methods in the implementation to make things work.

ok

[czgdp1807]

May be something using, self.front and self.rear. What's there in the parent classes? What's the MRO?

[HarsheetKakar]

tried using front and rear but this worked better since it is circular front can be behind rear in 2 cases if the deque is empty or if it is full and since % operator is used there is no way to find which one is the case

[czgdp1807]

+1

[czgdp1807]

In queue, elements are accessed from end points so these methods shouldn't be defined.

[HarsheetKakar]

helps a lot in rest of the implementation since I dont have to do modular arithmetic every time I push or pop

[czgdp1807]

At least they shouldn't be public. You can define helper methods, though 2 line is not much for code indirection.

[HarsheetKakar]

and these functions are very helpful if u want to change the value of some key in the queue, the user does not have to think about rear and front they can just start from 0 and go up to self.size. It actually makes the code a lot cleaner.

[czgdp1807]

the user does not have to think about rear and front

If someone wants to change the value of a key(somewhere other than front and rear), they could have used some other data structure.

[HarsheetKakar]

if we dont override it here then if user uses the index it will use DODA's getitem and setitem which will break the logic, and anyway it isnt breaking the already set datastructures and it helps a lot in development also as you can see in append/ appendleft. It makes the code readable, it helps to remove unnecessary bugs and there are only 2 small methods in question.

[HarsheetKakar]

and sir it also helps other developers to understand the code.

[czgdp1807]

Can we do the following,

class ArrayDeque(ArrayQueue):
   ...

[HarsheetKakar]

Can we do the following,

class ArrayDeque(ArrayQueue):
   ...

Error while trying to put it as ArrayQueue
Exception has occurred: TypeError multiple bases have instance lay-out conflict

[HarsheetKakar]

@czgdp1807 shall I add LinkedList implementation as well since then both of them would come under Deque abstract class and there would be no inconsistency with the already written code?

[czgdp1807]

I think the classes, ArrayDeque use DODA rather than inheriting and overriding it's methods. You can add LinkedList implementation, though again it should be used as a internal data member rather than being inherited.

[HarsheetKakar]

I think the classes, ArrayDeque use DODA rather than inheriting and overriding it's methods. You can add LinkedList implementation, though again it should be used as a internal data member rather than being inherited.

the append and pop method in ArrayDeque is quite different from DODA so if I have to use its properties and not its caveats I have to inherit it. I can also implement it from scratch but it will take more time.

[czgdp1807]

Any news?

[HarsheetKakar]

Any news?

sorry forgot about it, give me a couple of days

[HarsheetKakar]

I think the classes, ArrayDeque use DODA rather than inheriting and overriding it's methods. You can add LinkedList implementation, though again it should be used as a internal data member rather than being inherited.

the append and pop method in ArrayDeque is quite different from DODA so if I have to use its properties and not its caveats I have to inherit it. I can also implement it from scratch but it will take more time.

and you havent answered this one.

[czgdp1807]

Can we do append and pop as follows,

def append(self, x):
    self.queue.append(x)
    self.rear += 1

def pop(self, x):
    self.queue.delete(front)
    self.front += 1
    # check corner cases like if it's empty

[HarsheetKakar]

Can we do append and pop as follows,

def append(self, x):
    self.queue.append(x)
    self.rear += 1

def pop(self, x):
    self.queue.delete(front)
    self.front += 1
    # check corner cases like if it's empty

I dont think we can since rear is moved back to 0 if the array is doubled, this helps in keeping the _modify method less complicated.

[czgdp1807]

Probably, keeping, rear as self._last_pos_filled and tracking front by checking when the array will be double should do the the thing.

[HarsheetKakar]

Probably, keeping, rear as self._last_pos_filled and tracking front by checking when the array will be double should do the the thing.

sir but it is doubly ended so that means that rear also would move in the opposite direction, keeping track of both is important... with that said I have forgotten the implementation of most of it and would like to work on some other issue, so if its not a problem with you, we can put upforgrab on it... since most of the work is done someone can refactor the code easily

[czgdp1807]

Closing in favor of #321