All 15 Comprehensions & Generators questions
More in Comprehensions & Generators
How do you write a nested comprehension, like flattening a list of lists into a single flat list? How do you send a value INTO a generator using the .send() method, and how does this differ from just calling next()? How does 'yield from' delegate iteration to a sub-generator, and what does it return as its own expression value? How does generator.throw() let you inject an exception into a paused generator, and how would the generator handle it? How would you implement a memory-efficient sliding window or batching generator over a large iterable? How do you correctly close a generator early and ensure any try/finally cleanup inside it still runs? How would you implement a coroutine-style generator-based pipeline (the classic pre-async/await pattern) using multiple chained 'send()'-driven generators? How do you write a comprehension with MULTIPLE 'if' conditions, and how does that differ from combining them with 'and'? How would you profile whether a generator expression is actually providing a measurable memory or performance benefit over a list comprehension in a specific use case?
Advertisement
Python › Comprehensions & Generators

How does lazy evaluation in a chained sequence of generator expressions actually process elements, one at a time end-to-end, without intermediate lists?

Advanced

Answer

Each generator expression only pulls from its SOURCE generator when IT is asked for its own next value, propagating the request all the way back to the original iterable — meaning a chain of generator expressions processes ONE element completely through the entire pipeline before moving to the next, rather than processing each stage across the whole dataset before the next stage begins.
data = range(1_000_000)
pipeline = (x * 2 for x in (y + 1 for y in data if y % 2 == 0))
first_five = [next(pipeline) for _ in range(5)]
# Only the first ~10 elements of 'data' were ever actually touched, not the whole million
Real-world example Processing a huge dataset through a multi-stage transformation pipeline while only ever holding one element's worth of state at each stage.

Common follow-ups: How does this pull-based, element-at-a-time evaluation order differ from how a similarly-chained list comprehension would behave?

Memory Management & Garbage Collection
All 15 Comprehensions & Generators questions