Juan Pedro Bolivar Puente explains immutable and persistent data structures in C++ and how they support value semantics. He presents the Immer library, rrb-trees, and a prototype text editor that opens a huge file instantly.
Why Hacker News points people here: HN people point to it for rrb-trees, text editor data structures and bringing Clojure-style persistent structures to C++.
“Immer [0][1] seems like a really serious attempt at bringing high-quality persistent data structures to C++.”HN comment, 2021
“Juan Pedro is a next level programmer who I think has pushed the craft forward”
Better undo pattern than Command Pattern from Ableton developer
Full comment
Hey. Check out Juan Pedro Bolivar Puente's talks om value oriented design. He worked on Ableton which is a large application. He specifically speaks to Undo. He describes a much better pattern than the Command Pattern. He also released his source for Value Oriented Data structures Juan Pedro is a next level programmer who I think has pushed the craft forward through his work on bringing Value Oriented Design into the C++ world. Talk I this video Talk II link Data Structures: link Application Structure: link Text Editor: link Good luck.
RecommendsproverbialbunnyAug 2019
“talks about rrb-trees and how they're a great data structure for writing a text editor that can handle unlimited sized documents without locking up”
Using rrb-trees for large-document text editing with undo
Full comment
In case anyone hasn't seen it this CppCon talk this video talks about rrb-trees and how they're a great data structure for writing a text editor that can handle unlimited sized documents without locking up, and allows for editing while saving and opening, and an easy way to support an undo command. Most text editors today do not support such functionality. With lockless data structures like an rrb-tree it becomes easy. It really is an ideal way to do write a text editor.
RecommendsproverbialbunnySep 2019
“Not a benchmark but rrb-trees are pretty neat.”
Fun look at rrb-trees
Full comment
Not a benchmark but rrb-trees are pretty neat. They give an effective O(1) for all operations. Modern languages like Scala's Vector type is an rrb-tree under the hood: link For more fun: this video
RecommendsproverbialbunnyJun 2020
“If anyone wants a backstory building up to this data structure checkout this talk”
Backstory building up to an rrb-tree
Full comment
Mmm tasty. If anyone wants a backstory building up to this data structure checkout this talk: this video He builds an rrb-tree which uses the same underlying concepts ALEX uses, just without the ML-enhancement.
RecommendsbertmuthalalySep 2021
“Immer [0][1] seems like a really serious attempt at bringing high-quality persistent data structures to C++.”
Introduces Immer persistent data structure library for C++
Full comment
Immer [0][1] seems like a really serious attempt at bringing high-quality persistent data structures to C++. [0]: this video [1]: link
RecommendsmodernerdJun 2023
“See the author instantly opening a ~1GB text file with async loading”
Prototype editor opening a 1GB file instantly using immutable vectors
Full comment
You might be interested in ewig and immer by Juan Pedro Bolivar Puente: linklink See the author instantly opening a ~1GB text file with async loading, paging through, copying/pasting, and undoing/redoing in their prototype “ewig” text editor about 27 minutes into their talk here: this video It’s backed by a “vector of vectors” data structure called a relaxed radix balanced tree: link That original paper has seen lots of attention and attempts at performance improvements, such as: linklink
RecommendszelphirkaltOct 2025
“an impressive talk, in which someone shows their developed editor that can edit huge text files without lag”
Editor handling huge files with functional data structures
Full comment
> [...] which is a must for anything performance oriented. But that's also not really true. Using functional data structures it is possible to write highly performant software. Check out for example "CppCon 2017: Juan Pedro Bolivar Puente - Postmodern immutable data structures"[1], an impressive talk, in which someone shows their developed editor that can edit huge text files without lag faster than any mainstream editor today. This is enabled by functional data structures. Usage of functional data structures throughout a project also ensures, that parts can be trivially parallelized, which is not true for traditional algorithms, and therefore can easily scale with number of CPUs available. For more on that subject, you might want to watch a few Joe Armstrong talks. Rust achieves type safety by introducing more language concepts like ownership, borrowing, and lifetimes, to formalize, who has access when and how. It also creates things as immutable by default, unless you make use of `mut`. Unfortunately, Rust's philosophy is shared memory and making shared memory access safe, rather than message passing, which is at odds with FP. [1]: this video
Cites itarximboldiNov 2017
“My most popular open-source project [1] I talk about at conferences, mostly C++ ones”
Author's own conference talk about his open-source library
Full comment
My most popular open-source project [1] I talk about at conferences, mostly C++ ones (last example [2]). I was lucky to also land a paper about it at ICFP'17 [3]. On Friday I will talk at the MeetingCpp conference in Berlin about a new library I am building (kind of experimental Redux for C++ with time-travelling debugger!). I work as a freelancer and at the moment I do client work 3 days a week and dedicate the rest of the time to the open-source work. The open-source work includes doing research and implementing these tools, but also preparing the talks, which is a very time consuming task for me. Doing conference talks is very emotionally exhausting, I dread the previous weeks to the talk---but the feeling afterwards is very good and it is a good way to get exposure, I hope. My dream is to some day form a social company with a coop structure that works full-time on open-source, either directly or indirectly helping clients use these tools in their system. I hope I am set up in the right direction. Recently I opened a Twitter account to talk mostly about my open-source stuff [1]. I admittedly hate Twitter as a social network, but I've been convinced that it is a good place for networking in our industry, so I just play the game. I keep telling myself that I should open blog too, but then I realize that one can't simply do everything all the time, so I guess maybe I'll never do it. To your last question: is it unfair to not do marketing to your work? Totally not. Just do wha
Cites itarximboldiOct 2018
“my goals is to make C++ more like Clojure”
Bringing Clojure-style persistent data structures to C++
Full comment
Interesting to hear about C++ in this context, since of my goals is to make C++ more like Clojure: this videolink There is a lot of I could say about the relationship between C++ and Clojure, but I'll leave it at that for now :-)
Cites itproverbialbunnyAug 2019
“this cppcon talk is introduces one, which gives near O(1) time for all uses”
Example of a powerful functional data structure
Full comment
One example is streaming. Streams allow for a near infinite array size to be processed, because it works on one element at a time as it comes in. Streams come from the FP world. From that function chaining originates from the FP world and is used in modern languages like Python quite heavily. You mentioned research, so a valid example of this is the difference between Spark and MapReduce. Spark was introduced to optimize processing large sets of data, by streaming it. MapReduce does not. For a more heavy example, a few data structures that are quite powerful come from the FP world. For example, this cppcon talk is introduces one, which gives near O(1) time for all uses (insert, delete, lookup, ...), and can be used in a thread safe way: this video
Cites itproverbialbunnySep 2019
“If interested about the data structure”
Explains the rrb-tree data structure
Full comment
Interesting. I wonder how similar a Judy array is to an rrb-tree (relaxed radix binary tree), or if they're the same thing. An rrb-tree is an effective O(1) time for every category. Scala's Vector type implements one. If interested about the data structure: this video Scala bigO doc: link Scala's doc on it: link
Cites itarximboldiJan 2021
“It's design is covered in these talks”
Talks covering the design of a didactic text editor
Full comment
For another approach: I built a didactic text editor to teach "value oriented design" and immutable data-structures in C++: link It's design is covered in these talks: - Postmodern immutable data structures: this video - The most valuable values: link
Cites ithenrydarkJan 2023
“A nice showcase of this idea is in Juan Pedro Bolivar Puente's CppCon '17 talk”
Shows immutable structures need not copy all state
Full comment
Not to disagree, but to be fair modern fp runtimes use data structures that wouldn't have to copy so many bytes as the whole state. A nice showcase of this idea is in Juan Pedro Bolivar Puente's CppCon '17 talk Postmodern Immutable Data Structures: this video
