I have a type like this:
type TaskRow =
{
RowIndex : int
TaskId : string
Task : Task option
}
A function returns a list of these records to be processed further. Some of the functions doing that processing are only relevant for TaskRow items where Task is Some. I'm wondering what the best way is to go about that.
The naive way would be doing
let taskRowsWithTasks = taskRows |> Seq.filter (fun row -> Option.isSome row.Task)
and passing that to those functions, simply assuming that Task will never be None and using Task.Value, risking an NRE if I don't pass in that one special list. That is exactly what the current C# code does but seems rather unidiomatic for F#. I shouldn't be 'assuming' things but rather let the compiler tell me what will work.
More 'functional' would be to pattern match every time the value is relevant and then do/return nothing (and use choose or the like) for None, but that seems repetitive and wasteful as the same work would be done multiple times.
Another thought was introducing a second, slightly different type:
type TaskRowWithTask =
{
RowIndex : int
TaskId : string
Task : Task
}
The original list would then be filtered into a 'sublist' of this type one to be used where appropriate. I guess that would be okay from a functional perspective, but I wonder whether there's a nicer, idiomatic way without resorting to this kind of 'helper type'.
Thanks for any pointers!
There's quite a bit of value knowing that the tasks have already been filtered, so having two different types can be helpful. Instead of defining two different types (which, in F#, isn't that big a deal, though), you could also consider defining a generic Row type:
type Row<'a> = {
RowIndex : int
TaskId : string
Item : 'a }
This enables you to define a projection like this:
let project = function
| { RowIndex = ridx; TaskId = tid; Item = Some t } ->
Some { RowIndex = ridx; TaskId = tid; Item = t }
| _ -> None
let taskRowsWithTasks =
taskRows
|> Seq.map project
|> Seq.choose id
If the initial taskRows value has the type seq<Row<Task option>>, then the resulting taskRowsWithTasks sequence has the type seq<Row<Task>>.
I agree with you, the more "pure functional" way is to repeat the pattern match, I mean use a function with Seq.choose that does the filtering, instead of saving it to another structure.
let tasks = Seq.choose (fun {Task = t} -> t) taskRows
The problem is performance as it would be calculated many times, but you can use Seq.cache so behind the scenes it's saved into an intermediate structure, while keeping your code more "pure functional" looking.
Related
I am not a functional programmer.
I am learning F#.
I got a problem here.
Let me start from following piece of code:
type XmlNode(tagName, innerValue) =
member this.TagName = tagName
member this.InnerValue = innerValue
member this.Atts = Dictionary<string, obj>()
I don't use F# dict because (as I know) that one is readonly, however I obviously need to modify my attributes.
So I am really struggling to make it pure functional way:
type XmlNode with member this.WriteTo (output:StringBuilder) =
output.Append("<" + this.TagName) |> ignore
//let writeAtts =
// List.map2 (fun key value -> " " + key + "=" + value.ToString())
(List.ofSeq this.Atts.Keys) (List.ofSeq this.Atts.Values)
// |> List.reduce (fun acc str -> acc + " " + str)
//output.Append((writeAtts)) |> ignore
output.Append(">" + this.InnerValue + "</" + this.TagName + ">") |> ignore
output
The code I commented out was my (probably stupid) attemp to use mapping and reduction to concat all the atts in the single correctly formatted string. And that compiles OK.
But when I try to access my Atts property:
[<EntryPoint>]
let main argv =
let root = new XmlNode("root", "test")
root.Atts.Add("att", "val") // trying to add a new KVP
let output = new StringBuilder()
printfn "%O" (root.WriteTo(output))
Console.ReadLine()|>ignore
0 // return an integer exit code
...new attribute does not appear inside the Atts property, i.e. it remains empty.
So:
1) help me to make my code more functional.
2) and to understand how to deal with modificable dictionaries in F#.
Thank you.
First, your immediate problem: the way you defined the Atts property, it's not one value that is "stored" somewhere and is accessible via property. Instead, your definition means "every time somebody reads this property, create a new dictionary and return it". This is why your new attribute doesn't appear in the dictionary: it's a different dictionary every time you read root.Atts.
To create a property with a backing field and initial value, use member val:
type XmlNode(...) =
...
member val Atts = Dictionary<string,object>()
Now, answers to some implied questions.
First order of business: "modify the attributes" and "purely functional" are contradictory ideas. Functional programming implies immutable data. Nothing changes ever. The way to advance your computation is to create a new datum at every step, without overwriting the previous one. This basic idea turns out to be immensely valuable in practice: safer threading, trivial "undo" scenarios, trivial parallelization, trivial distribution to other machines, and even reduced memory consumption via persistent data structures.
Immutability is a very important point, and I urge you not to glance over it. Accepting it requires a mental shift. From my own (and other people I know) experience, it is very hard coming from imperative programming, but it is well worth it.
Second: do not use classes and properties. Technically speaking, object-oriented programming (in the sense of message passing) is not contradictory to functional, but the Enterprise flavor that is used in practice and implemented in C++, Java, C# et al., is contradictory, because it emphasizes this idea that "methods are operations that change an object's state", which is not functional (see above). So it's better to avoid object-oriented constructs, at least while you're learning. And especially since F# provides much better ways to encode data:
type XmlNode = { TagName: string; InnerValue: string; Atts: (string*string) list }
(notice how my Atts is not a dictionary; we'll come to this in a bit)
Similarly, to represent operations on your data, use functions, not methods:
let printNode (node: XmlNode) = (* we'll come to the implementation later *)
Third: why do you say that you "obviously" need to modify the attributes? The code you've shown does not call for this. For example, using my definition of XmlNode above, I can rewrite your code this way:
[<EntryPoint>]
let main argv =
let root = { TagName = "root"; InnerValue = "test"; Atts = ["att", "val"] }
printfn "%s" (printNode root)
...
But even if that was a real need, you shouldn't do it "in place". As I've described above while talking about immutability, you should not mutate the existing node, but rather create a new node that differs from the original one in whatever way you wanted to "modify":
let addAttr node name value = { node with Atts = (name, value) :: node.Atts }
In this implementation, I take a node and name/value of an attribute, and produce a new node whose Atts list consists of whatever was in the original node's Atts with the new attribute prepended.
The original Atts list stays intact, unmodified. But this does not mean twice the memory consumption: because we know that the original list never changes, we can reuse it: we create the new list by only allocating memory for the new item and including a reference to the old list as "other items". If the old list was subject to change, we couldn't do that, we would have to create a full copy (see "Defensive Copy"). This strategy is known as "Persistent Data Structure". It is one of the pillars of functional programming.
Finally, for string formatting, I recommend using sprintf instead of StringBuilder. It offers similar performance benefits, but in addition provides type safety. For example, code sprintf "%s" 5 will not compile, complaining that the format expects a string, but the final argument 5 is a number. With this, we can implement the printNode function:
let printNode (node: XmlNode) =
let atts = seq { for n, v in node.Atts -> sprintf " %s=\"%s\"" n v } |> String.concat ""
sprintf "<%s%s>%s</%s>" node.TagName atts node.InnerValue node.TagName
For reference, here's your complete program, rewritten in functional style:
type XmlNode = { TagName: string; InnerValue: string; Atts: (string*string) list }
let printNode (node: XmlNode) =
let atts = seq { for n, v in node.Atts -> sprintf " %s=\"%s\"" n v } |> String.concat ""
sprintf "<%s%s>%s</%s>" node.TagName atts node.InnerValue node.TagName
[<EntryPoint>]
let main argv =
let root = { TagName = "root"; InnerValue = "test"; Atts = ["att", "val"] }
printfn "%s" (printNode root)
Console.ReadLine() |> ignore
0
The Stats.expandingXXXX functions are pretty fast. However there is no public api to do a expandingWindow apply. The following solution i created is really slow when it comes to large dataset like 100k. Any suggestion is appreciated?
let ExpWindowApply f minSize data =
let keys = dataSeries.Keys
let startKey = dataSeries.FirstKey()
let values = keys
|> Seq.map(fun k ->
let ds = data.Between(startKey,k)
match ds with
|_ when ds.ValueCount >= minSize -> f ds.Values
|_ -> Double.NaN
)
let result = Series(keys, values)
result
I understand the Stats.expandingXXX function are actually special cases where the function being applied can be iterately calculated based on previous loop's state. And not all function can take advantage of states from previous calculation. Is there anything better way than Series.Between in terms of creating a window of data?
Update
For those who are also interested in the similar issue. The answer provides alternative implementation and insight into rarely documented series vector and index operation. But it doesn't improve performance.
The expanding functions in Deedle are fast because they are using an efficient online algorithm that makes it possible to calculate the statistics on the fly with just one pass - rather than actually building the intermediate series for the sub-ranges.
There is a built-in function aggregate that lets you do something this - though it works in the reversed way. For example, if you want to sum all elements starting from the current one to the end, you can write:
let s = series [ for i in 1 .. 10 -> i, float i ]
s |> Series.aggregateInto
(Aggregation.WindowWhile(fun _ _ -> true))
(fun seg -> seg.Data.FirstKey())
(fun seg -> OptionalValue(Stats.sum seg.Data))
If you want to do the same thing using the underlying representation, you can directly use the addressing scheme that Deedle uses to link the keys (in the index) with values (in the data vector). This is an ugly mutable sample, but you can encapsulate it into something nicer:
[ let firstAddr = s.Index.Locate(s.FirstKey())
for k in s.Index.KeySequence ->
let lastAddr = s.Index.Locate(k)
seq {
let a = ref firstAddr
while !a <> lastAddr do
yield s.Vector.GetValue(!a).Value
a := s.Index.AddressOperations.AdjustBy(!a, +1L) } |> Seq.sum ]
I need a data structure for the following:
In a device that has memory slots, each of the slots has a set of parameters. These parameters have different types. The list of possible parameters is fixed, so there is no need for generic flexibility à la »Support of arbitrary parameters without change«. Also, for each parameter, the structure of the contents is known. Typical use cases are the retrieval and modification of one specific parameter as well as a transformation of the complete parameter set into a different (but already defined) data structure.
The natural choice of F# data structure would be a sum type like this:
type SomeParameterContentType = { Field1 : string, Field2 : int }
type SomeOtherParameterContentType = { Other1 : bool option, Other2 : double }
type Parameter =
| SomeParameter of SomeParameterContentType
| SomeOtherParameter of SomeOtherParameterContentType
This way I could create a set and store the parameters there with a very nice data structure. The question here is: Given this idea, how would looking for a specific parameter look like? I don't know of any way to specify a predicate for a find-function for sets. It would be possible to define another sum type listing just the Parameter Types without their contents using this as key for a Dictionary but I don't like this idea too much. Using strings instead of the second sum type doesn't make things better as it still would require providing the list of possible parameters twice.
Does anyone have a better idea?
Thx
--Mathias.
Sounds like all you want is a tryFind for a Set:
module Set =
let tryFind p =
Set.toList >> List.tryFind p
Usage:
let mySet = Set.ofList [1;2;3;4;5]
let m = mySet |> Set.tryFind (fun t -> t = 2)
val m : int option = Some 2
Usage with your Types:
let yourSet = Set.ofList [SomeParameter {Field1="hello";Field2=3}]
let mYours = yourSet |> Set.tryFind (fun t -> match t with
|SomeParameter p -> true
|SomeOtherParameter p -> false)
val mYours : Parameter option = Some (SomeParameter {Field1 = "hello";
Field2 = 3;})
I have written a simple example for my scenario. I create a record type Switch
type State =
| On
| Off
with
member this.flip =
match this with
| On -> Off
| Off -> On
type Switch = { State : State }
and then I write a function that creates a copy of the record with one element changed
let flip switch = { switch with State = switch.State.flip }
To flip many successive times I write
let flipMany times switch =
[1 .. times]
|> List.fold (fun (sw : Switch) _ -> flip sw) switch
If I want put these two functions on the record as methods, I write instead
type Switch =
{ State : State }
member this.flip =
{ this with State = this.State.flip }
member this.flipMany times =
[1 .. times]
|> List.fold (fun (sw : Switch) _ -> sw.flip) this
Is there anything wrong with doing this? Is it equally efficient? It feels a bit uncomfortable calling the function sw.flip on a different object every single time.
Edit: this is just a simple example in order to explain my question. My question is on how the function flipMany compares with the flipMany method on the record. The implementation might be naive, but it is the same in both of cases.
Your intent can be implemented as simple as
let flipMany times switch =
match (times % 2) with
| 1 -> { switch with State = switch.State.flip }
| _ -> switch
type Switch =
{ State : State }
member this.Flip = { this with State = this.State.flip }
member this.FlipMany times =
match (times % 2) with | 1 -> this.Flip | _ -> this
In the broader context of comparing a static function versus an object's method the idiomatic way would be sticking to function option. A function has explicit arguments and must not depend on any side state, but state of arguments, to produce an idempotent result value. On the contrary, an object method implicitly gets an instance of the class as an argument and may derive result value not just from arguments, but based on the state of other class fields too, which is not in line with idempotency property of a pure function.
To feel this difference better it may help reading thru F# Components Design Guidelines and exploring F# Core libraries design.
I am writing a compiler of mini-pascal in Ocaml. I would like my compiler to accept the following code for instance:
program test;
var
a,b : boolean;
n : integer;
begin
...
end.
I have difficulties in dealing with the declaration of variables (the part following var). At the moment, the type of variables is defined like this in sib_syntax.ml:
type s_var =
{ s_var_name: string;
s_var_type: s_type;
s_var_uniqueId: s_uniqueId (* key *) }
Where s_var_uniqueId (instead of s_var_name) is the unique key of the variables. My first question is, where and how I could implement the mechanism of generating a new id (actually by increasing the biggest id by 1) every time I have got a new variable. I am wondering if I should implement it in sib_parser.mly, which probably involves a static variable cur_id and the modification of the part of binding, again don't know how to realize them in .mly. Or should I implement the mechanism at the next stage - the interpreter.ml? but in this case, the question is how to make the .mly consistent with the type s_var, what s_var_uniqueId should I provide in the part of binding?
Another question is about this part of statement in .mly:
id = IDENT COLONEQ e = expression
{ Sc_assign (Sle_var {s_var_name = id; s_var_type = St_void}, e) }
Here, I also need to provide the next level (the interpreter.ml) a variable of which I only know the s_var_name, so what could I do regarding its s_var_type and s_var_uniqueId here?
Could anyone help? Thank you very much!
The first question to ask yourself is whether you actually need an unique id. From my experience, they're almost never necessary or even useful. If what you're trying to do is making variables unique through alpha-equivalence, then this should happen after parsing is complete, and will probably involve some form of DeBruijn indices instead of unique identifiers.
Either way, a function which returns a new integer identifier every time it is called is:
let unique =
let last = ref 0 in
fun () -> incr last ; !last
let one = unique () (* 1 *)
let two = unique () (* 2 *)
So, you can simply assign { ... ; s_var_uniqueId = unique () } in your Menhir rules.
The more important problem you're trying to solve here is that of variable binding. Variable x is defined in one location and used in another, and you need to determine that it happens to be the same variable in both places. There are many ways of doing this, one of them being to delay the binding until the interpreter. I'm going to show you how to deal with this during parsing.
First, I'm going to define a context: it's a set of variables that allows you to easily retrieve a variable based on its name. You might want to create it with hash tables or maps, but to keep things simple I will be using List.assoc here.
type s_context = {
s_ctx_parent : s_context option ;
s_ctx_bindings : (string * (int * s_type)) list ;
s_ctx_size : int ;
}
let empty_context parent = {
s_ctx_parent = parent ;
s_ctx_bindings = [] ;
s_ctx_size = 0
}
let bind v_name v_type ctx =
try let _ = List.assoc ctx.s_ctx_bindings v_name in
failwith "Variable is already defined"
with Not_found ->
{ ctx with
s_ctx_bindings = (v_name, (ctx.s_ctx_size, v_type))
:: ctx.s_ctx_bindings ;
s_ctx_size = ctx.s_ctx_size + 1 }
let rec find v_name ctx =
try 0, List.assoc ctx.s_ctx_bindings v_name
with Not_found ->
match ctx.s_ctx_parent with
| Some parent -> let depth, found = find v_name parent in
depth + 1, found
| None -> failwith "Variable is not defined"
So, bind adds a new variable to the current context, find looks for a variable in the current context and its parents, and returns both the bound data and the depth at which it was found. So, you could have all global variables in one context, then all parameters of a function in another context that has the global context as its parent, then all local variables in a function (when you'll have them) in a third context that has the function's main context as the parent, and so on.
So, for instance, find 'x' ctx will return something like 0, (3, St_int) where 0 is the DeBruijn index of the variable, 3 is the position of the variable in the context identified by the DeBruijn index, and St_int is the type.
type s_var = {
s_var_deBruijn: int;
s_var_type: s_type;
s_var_pos: int
}
let find v_name ctx =
let deBruijn, (pos, typ) = find v_name ctx in
{ s_var_deBruijn = deBruijn ;
s_var_type = typ ;
s_var_pos = pos }
Of course, you need your functions to store their context, and make sure that the first argument is the variable at position 0 within the context:
type s_fun =
{ s_fun_name: string;
s_fun_type: s_type;
s_fun_params: context;
s_fun_body: s_block; }
let context_of_paramlist parent paramlist =
List.fold_left
(fun ctx (v_name,v_type) -> bind v_name v_type ctx)
(empty_context parent)
paramlist
Then, you can change your parser to take into account the context. The trick is that instead of returning an object representing part of your AST, most of your rules will return a function that takes a context as an argument and returns an AST node.
For instance:
int_expression:
(* Constant : ignore the context *)
| c = INT { fun _ -> Se_const (Sc_int c) }
(* Variable : look for the variable inside the contex *)
| id = IDENT { fun ctx -> Se_var (find id ctx) }
(* Subexpressions : pass the context to both *)
| e1 = int_expression o = operator e2 = int_expression
{ fun ctx -> Se_binary (o, e1 ctx, e2 ctx) }
;
So, you simply propagate the context "down" recursively through the expressions. The only clever parts are those when new contexts are created (you don't have this syntax yet, so I'm just adding a placeholder):
| function_definition_expression (args, body)
{ fun ctx -> let ctx = context_of_paramlist (Some ctx) args in
{ s_fun_params = ctx ;
s_fun_body = body ctx } }
As well as the global context (the program rule itself does not return a function, but the block rule does, and so a context is created from the globals and provided).
prog:
PROGRAM IDENT SEMICOLON
globals = variables
main = block
DOT
{ let ctx = context_of_paramlist None globals in
{ globals = ctx;
main = main ctx } }
All of this makes the implementation of your interpreter much easier due to the DeBruijn indices: you can have a "stack" which holds your values (of type value) defined as:
type stack = value array list
Then, reading and writing variable x is as simple as:
let read stack x =
(List.nth stack x.s_var_deBruijn).(x.s_var_pos)
let write stack x value =
(List.nth stack x.s_var_deBruijn).(x.s_var_pos) <- value
Also, since we made sure that function parameters are in the same order as their position in the function context, if you want to call function f and its arguments are stored in the array args, then constructing the stack is as simple as:
let inner_stack = args :: stack in
(* Evaluate f.s_fun_body with inner_stack here *)
But I'm sure you'll have a lot more questions to ask when you start working on your interpeter ;)
How to create a global id generator:
let unique =
let counter = ref (-1) in
fun () -> incr counter; !counter
Test:
# unique ();;
- : int = 0
# unique ();;
- : int = 1
Regarding your more general design question: it seems that your data representation does not faithfully represent the compiler phases. If you must return a type-aware data-type (with this field s_var_type) after the parsing phase, something is wrong. You have two choices:
devise a more precise data representation for the post-parsing AST, that would be different from the post-typing AST, and not have those s_var_type fields. Typing would then be a conversion from the untyped to the typed AST. This is a clean solution that I would recommend.
admit that you must break the data representation semantics because you don't have enough information at this stage, and try to be at peace with the idea of returning garbage such as St_void after the parsing phase, to reconstruct the correct information later. This is less typed (as you have an implicit assumption on your data which is not apparent in the type), more pragmatic, ugly but sometimes necessary. I don't think it's the right decision in this case, but you will encounter situation where it's better to be a bit less typed.
I think the specific choice of unique id handling design depends on your position on this more general question, and your concrete decisions about types. If you choose a finer-typed representation of post-parsing AST, it's your choice to decide whether to include unique ids or not (I would, because generating a unique ID is dead simple and doesn't need a separate pass, and I would rather slightly complexify the grammar productions than the typing phase). If you choose to hack the type field with a dummy value, it's also reasonable to do that for variable ids if you wish to, putting 0 as a dummy value and defining it later; but still I personally would do that in the parsing phase.