Re: endsWith - for a string vs an array of strings
On Saturday, 10 January 2015 at 23:32:47 UTC, bearophile wrote: Laeeth Isharc: In D there is a feature that allows a function to accept both an array of items and items, yes - it is funny there is not an overloading that accepts arrays I meant this D feature: void foo(T)(T[] items...) { import std.stdio; items.writeln; } void main() { foo("red", "green", "blue"); foo(["red", "green", "blue"]); auto a = ["red", "green", "blue"]; foo(a); } Bye, bearophile for the curious, expanding tuples and typetuples on ali's book is explained at http://ddili.org/ders/d.en/tuples.html#ix_tuples..expand and at http://ddili.org/ders/d.en/tuples.html#ix_tuples.TypeTuple, std.typetuple.
Re: Sqlite
On Sunday, 11 January 2015 at 22:19:28 UTC, Tobias Pankrath wrote: Hint: Put the SQL in a file create_people.sql and import it into your code via the import statement: string sql = import("create_people.sql"); // you'll need a correct -J compiler switch That way you can easily test if it's correct on the commandline. It's .read in the sqlite3 shell. Neat, thank you!
Re: How to interface with C++ code or dll?
On Sunday, 11 January 2015 at 12:56:40 UTC, Suliman wrote: I had read about using C++ libs from D, and understood that there is 2 ways: 1. Make binding - convert .H to .d (I still do not fully understand what is it) 2. Use directly C++ lib (.dll) No, there are not two ways. There is one way. In order to use any library from C or C++, you *have* to create a binding. The binding tells the compiler the function signatures and types that are available to be used. It allows you to compile your code without error. Then, the C or C++ library can be linked with the final executable just like a D library, or (in the case of shared libraries) loaded at runtime (as Derelict does) with system calls. I decided to start from simple example http://www.gdal.org/warptut.html For compiling it's need #include "gdalwarper.h", so I maked gdalwarper.d with tool htod. I know that it's not recommended for usage more, but other tools look too complex for me (I even do not understand how to sun them on Windows). After I had add import gdalwarper.d to my test project. But during compilation it's asked me about other d files, that I included one ny other in App.d file. In the result I got error that: "source\cpl_port.d(147): Error: module ctype is in file 'std\c\ctype.d' which can not be read" You've imported std.c.ctype, which doesn't exist. You need to import core.stdc.ctype instead. All of the C stuff is in core.stdc in DRuntime. I have got a question. Files that I include is only header files. They do not include any code. How App would run? I should specify C++ source location or what? Imports are only used at compile time to determine which types and functions are available to use. We don't need the code for the C or C++ stuff because everything is already compiled in the library that you are linking to or loading. Even in D, you don't need the code for modules you import -- only their function signatures and type declarations. D is designed such that you don't need to separate the interface from the implementation, but you can if you want to (that's what .di files are for). To be clear, let's say you have module a.d and another module b.d. Module a has a function sayHello. You compile module a and the code for say hello, which prints hello to the screen, is translated and saved in a library. Later, you compile b.d. Since module b imports a, then the compiler has to load a into memory and parse the signature for the function sayHello -- but it doesn't *need* the code for say hello. All it needs is this part: void sayHello( string str ); It doesn't care what the implementation of the function looks like, because now it isn't compiling a.d, it's compiling b.d. It only needs to know how to call the function and what it returns so it can throw an error if b.d calls sayHello with an int or a float or anything that isn't a string. Then, when the executable is generated, the linker will match up the call to sayHello in b with the implementation of sayHello in a. There are some exceptions to this. You need the implementation of templates and anything you want to use at compile time, but for code that is runtime only you don't. = Then I decided to look how to use ready dll. As far as I know D have directive pragma(), that allow to import any lib. Also afaik that it support inly import of dll in format of lib. Am I right? No. This is not a feature of D. D has a "lib" pragma that allows you to link to a library without passing the library on the command line. THat is: pragma( lib, "MyLib.lib" ); This will cause the compiler to automatically link MyLib.lib to your app. It's the same as passing MyLib.lib on the command line. The compiler still needs to know the path where the library can be found. However, this does *not* allow you to automatically use any library in D without importing the library's source modules. So for a C or C++ library, you still need a binding to import.
Re: Sqlite
On Sunday, 11 January 2015 at 20:30:41 UTC, Paul wrote: On Sunday, 11 January 2015 at 20:20:21 UTC, ketmar via Digitalmars-d-learn wrote: note the single quotes in your code: that is where it all goes wrong. i don't know where you got that quotes from, but this is not a valid SQL syntax for `CREATE TABLE`. ;-) Thank you, I thought it might be something obvious - that will teach me to cut and paste code! :D Hint: Put the SQL in a file create_people.sql and import it into your code via the import statement: string sql = import("create_people.sql"); // you'll need a correct -J compiler switch That way you can easily test if it's correct on the commandline. It's .read in the sqlite3 shell.
Re: Build all combinations of strings
Nordlöw: Couldn't we do a first pass and check that if elements of T are distinct and if so set doCopy to false otherwise true? The algorithm you have seen in Rosettacode doesn't care if and what items of the input sequence are duplicated, it handles them as they are all distinct. And them being distinct (or not distinct) doesn't change the desire to use something like doCopy to have dup-ped output arrays, so I don't understand what you are trying to say. The purpose of doCopy is similar of the difference between File.byLine and File.byLineCopy (originally I suggested to give a doCopy argiment to byLine too, for safety. Andrei said no. Later experience has shown I was right and we have added byLineCopy, but now the default line iteration is the non-copying one, that is less safe). Bye, bearophile
Re: Build all combinations of strings
On Sunday, 11 January 2015 at 18:01:09 UTC, bearophile wrote: Is doCopy really needed as an argument here? Couldn't this be inferred from the mutability of T instead? doCopy is useful, if it's true all the permutation arrays are distinct and dup-ped, otherwise they are all different. It's true by default, so casual users of that generator will avoid bugs. You can set it to false to speed up your code. Couldn't we do a first pass and check that if elements of T are distinct and if so set doCopy to false otherwise true?
Re: Sqlite
On Sunday, 11 January 2015 at 20:20:21 UTC, ketmar via Digitalmars-d-learn wrote: note the single quotes in your code: that is where it all goes wrong. i don't know where you got that quotes from, but this is not a valid SQL syntax for `CREATE TABLE`. ;-) Thank you, I thought it might be something obvious - that will teach me to cut and paste code! :D
Re: Build all combinations of strings
On Sunday, 11 January 2015 at 18:01:09 UTC, bearophile wrote: Nordlöw: Is doCopy really needed as an argument here? Couldn't this be inferred from the mutability of T instead? doCopy is useful, if it's true all the permutation arrays are distinct and dup-ped, otherwise they are all different. It's true by default, so casual users of that generator will avoid bugs. You can set it to false to speed up your code. Later I have refined the idea, you can see it here, that allows true @nogc code when needed: struct CartesianPower(bool doCopy=true, T) { T[] items; uint repeat; T[] row; uint i, maxN; this(T[] items_, in uint repeat_, T[] buffer) pure nothrow @safe @nogc { this.items = items_; this.repeat = repeat_; row = buffer[0 .. repeat]; row[] = items[0]; maxN = items.length ^^ repeat; } static if (doCopy) { @property T[] front() pure nothrow @safe @nogc { return row.dup; } } else { @property T[] front() pure nothrow @safe @nogc { return row; } } @property bool empty() pure nothrow @safe @nogc { return i >= maxN; } void popFront() pure nothrow @safe @nogc { i++; if (empty) return; uint n = i; size_t count = repeat - 1; while (n) { row[count] = items[n % items.length]; count--; n /= items.length; } } } auto cartesianPower(bool doCopy=true, T)(T[] items, in uint repeat) pure nothrow @safe { return CartesianPower!(doCopy, T)(items, repeat, new T[repeat]); } auto cartesianPower(bool doCopy=true, T)(T[] items, in uint repeat, T[] buffer) pure nothrow @safe @nogc { if (buffer.length >= repeat) { return CartesianPower!(doCopy, T)(items, repeat, buffer); } else { // Is this correct in presence of chaining? static immutable err = new Error("buffer.length < repeat"); throw err; } } void main() @nogc { import core.stdc.stdio; int[3] items = [10, 20, 30]; int[4] buf; foreach (p; cartesianPower!false(items, 4, buf)) printf("(%d, %d, %d, %d)\n", p[0], p[1], p[2], p[3]); } Bye, bearophile Nice! PR anyone?
Re: Sqlite
On Sun, 11 Jan 2015 20:00:03 + Paul via Digitalmars-d-learn wrote: > Can someone please tell me what I'm doing wrong here, the sql > INSERT statement fails for some reason. I don't fully understand > the callback function yet (I borrowed this from a C tutorial on > the subject), maybe that is the source of the problem? > > > import etc.c.sqlite3; > import std.stdio; > > //stub > extern(C) int aCallback(void *n, int c, char **v, char **col) > { >return 0; > } > > void main(){ > > sqlite3 *db; > int result = sqlite3_open("myDatabase.db", &db); > > if (result) { > writeln("Failed to open database"); > return; > } > > //create table > char *msg = null; > result = sqlite3_exec(db, "CREATE TABLE people('id INT PRIMARY > KEY NOT NULL, surname TEXT NOT NULL');", &aCallback, null, &msg); > if (result) { > writeln("Failed to create table"); > > //tidy up on exit > sqlite3_close(db); > return; > } > > //insert record > msg = null; > result = sqlite3_exec(db, "INSERT INTO people (id, surname) > VALUES (1, 'Smith');", &aCallback, null, &msg); > if (result) { > writeln("Failed to insert record"); > > //tidy up on exit > sqlite3_close(db); > return; > > } > > sqlite3_free(msg); > sqlite3_close(db); > > } > > Many thanks > > Paul if you'll output the error message, you'll see something unusual here: if (result) { import std.conv : to; writeln("Failed to insert record: ", to!string(msg)); //tidy up on exit sqlite3_close(db); return; } "Failed to insert record: table people has no column named id" wow! but it has! or isn't it? yep, it hasn't. the error is here: > result = sqlite3_exec(db, "CREATE TABLE people('id INT PRIMARY > KEY NOT NULL, surname TEXT NOT NULL');", &aCallback, null, &msg); `CREATE TABLE people('...')` is not the syntax you want. i don't know why sqlite is not rejecting it, but the correct one is this: result = sqlite3_exec(db, "CREATE TABLE people(id INT PRIMARY "~ "KEY NOT NULL, surname TEXT NOT NULL);", &aCallback, null, &msg); note the single quotes in your code: that is where it all goes wrong. i don't know where you got that quotes from, but this is not a valid SQL syntax for `CREATE TABLE`. ;-) signature.asc Description: PGP signature
Sqlite
Can someone please tell me what I'm doing wrong here, the sql INSERT statement fails for some reason. I don't fully understand the callback function yet (I borrowed this from a C tutorial on the subject), maybe that is the source of the problem? import etc.c.sqlite3; import std.stdio; //stub extern(C) int aCallback(void *n, int c, char **v, char **col) { return 0; } void main(){ sqlite3 *db; int result = sqlite3_open("myDatabase.db", &db); if (result) { writeln("Failed to open database"); return; } //create table char *msg = null; result = sqlite3_exec(db, "CREATE TABLE people('id INT PRIMARY KEY NOT NULL, surname TEXT NOT NULL');", &aCallback, null, &msg); if (result) { writeln("Failed to create table"); //tidy up on exit sqlite3_close(db); return; } //insert record msg = null; result = sqlite3_exec(db, "INSERT INTO people (id, surname) VALUES (1, 'Smith');", &aCallback, null, &msg); if (result) { writeln("Failed to insert record"); //tidy up on exit sqlite3_close(db); return; } sqlite3_free(msg); sqlite3_close(db); } Many thanks Paul
Re: Build all combinations of strings
Nordlöw: Is doCopy really needed as an argument here? Couldn't this be inferred from the mutability of T instead? doCopy is useful, if it's true all the permutation arrays are distinct and dup-ped, otherwise they are all different. It's true by default, so casual users of that generator will avoid bugs. You can set it to false to speed up your code. Later I have refined the idea, you can see it here, that allows true @nogc code when needed: struct CartesianPower(bool doCopy=true, T) { T[] items; uint repeat; T[] row; uint i, maxN; this(T[] items_, in uint repeat_, T[] buffer) pure nothrow @safe @nogc { this.items = items_; this.repeat = repeat_; row = buffer[0 .. repeat]; row[] = items[0]; maxN = items.length ^^ repeat; } static if (doCopy) { @property T[] front() pure nothrow @safe @nogc { return row.dup; } } else { @property T[] front() pure nothrow @safe @nogc { return row; } } @property bool empty() pure nothrow @safe @nogc { return i >= maxN; } void popFront() pure nothrow @safe @nogc { i++; if (empty) return; uint n = i; size_t count = repeat - 1; while (n) { row[count] = items[n % items.length]; count--; n /= items.length; } } } auto cartesianPower(bool doCopy=true, T)(T[] items, in uint repeat) pure nothrow @safe { return CartesianPower!(doCopy, T)(items, repeat, new T[repeat]); } auto cartesianPower(bool doCopy=true, T)(T[] items, in uint repeat, T[] buffer) pure nothrow @safe @nogc { if (buffer.length >= repeat) { return CartesianPower!(doCopy, T)(items, repeat, buffer); } else { // Is this correct in presence of chaining? static immutable err = new Error("buffer.length < repeat"); throw err; } } void main() @nogc { import core.stdc.stdio; int[3] items = [10, 20, 30]; int[4] buf; foreach (p; cartesianPower!false(items, 4, buf)) printf("(%d, %d, %d, %d)\n", p[0], p[1], p[2], p[3]); } Bye, bearophile
Re: Build all combinations of strings
On Monday, 11 June 2012 at 19:52:38 UTC, bearophile wrote: Using that the code is: import std.string, std.stdio, std.array; void main() { auto words = "foo bar doo".split(); auto res = permutations!false(words).map!(p => p.join(" "))().array(); writeln(res); } Is doCopy really needed as an argument here? Couldn't this be inferred from the mutability of T instead?
How to interface with C++ code or dll?
I am trying to understand how to use C++ lib from D. For my App I need http://www.gdal.org I had read about using C++ libs from D, and understood that there is 2 ways: 1. Make binding - convert .H to .d (I still do not fully understand what is it) 2. Use directly C++ lib (.dll) I decided to start from simple example http://www.gdal.org/warptut.html For compiling it's need #include "gdalwarper.h", so I maked gdalwarper.d with tool htod. I know that it's not recommended for usage more, but other tools look too complex for me (I even do not understand how to sun them on Windows). After I had add import gdalwarper.d to my test project. But during compilation it's asked me about other d files, that I included one ny other in App.d file. In the result I got error that: "source\cpl_port.d(147): Error: module ctype is in file 'std\c\ctype.d' which can not be read" I have got a question. Files that I include is only header files. They do not include any code. How App would run? I should specify C++ source location or what? = Then I decided to look how to use ready dll. As far as I know D have directive pragma(), that allow to import any lib. Also afaik that it support inly import of dll in format of lib. Am I right? And next. How I can understand which lib do I need. All documentation of GDAL assume that programmer will use source code, but not libs. And what I should to do? I download gdal bins, and there is a lot of dlls and I even can't understand which of them I need.
Re: Tuple/Typedef question
I use this Typedef implementation instead: /// one with non-default initializer template Typedef(T, istring name, T initval) { static assert (name.length, "Can't create Typedef with an empty identifier"); enum Typedef = ("struct " ~ name ~ "{ " ~ T.stringof ~ " value = " ~ initval.stringof ~ ";" ~ "alias value this;" ~ " }"); } /// basic overload template Typedef(T, istring name) { static assert (name.length, "Can't create Typedef with an empty identifier"); enum Typedef = ("struct " ~ name ~ "{ " ~ T.stringof ~ " value; " ~ "alias value this;" ~ " }"); } unittest { mixin(Typedef!(int, "MyInt1")); mixin(Typedef!(int, "MyInt2")); static assert (!is(MyInt1 : MyInt2)); }
Re: Tuple/Typedef question
On Sun, 11 Jan 2015 12:00:19 + Martin via Digitalmars-d-learn wrote: > > as for `Typedef!` -- you can use it's third arg, "cookie": > > > > import std.typecons; > > > > alias T1 = Tuple!(int, int); > > alias T2 = Tuple!(int, int); > > > > alias T1New = Typedef!(T1, T1.init, "t0"); > > alias T2New = Typedef!(T2, T2.init, "t1"); > > > > pragma(msg, __traits(isSame, T1New, T2New)); // false > > > > there was a heated discussion about `std.typecons.Typedef`, > > built-in > > `typedef` and other related things, but the decision was to > > keep the > > status quo. > > I can't believe I missed the cookie part. Thanks! this part deserves a better explanation in docs, 'cause it's easy to miss the details, especially for those who assumed that each `Typedef!` creates a distinct type. too bad that my writing skills sux (and i'm really biased against the current `Typedef!`). signature.asc Description: PGP signature
Re: Tuple/Typedef question
On Sunday, 11 January 2015 at 11:52:42 UTC, ketmar via Digitalmars-d-learn wrote: On Sun, 11 Jan 2015 11:41:08 + Martin via Digitalmars-d-learn wrote: Is there a way to get Tuple (and Typedef) from the std.typecons module to generate a new type that is unique on every instantiation? What I mean is: alias T1 = Tuple!(int, int); alias T2 = Tuple!(int, int); writeln(__traits(isSame, T1, T2)); // prints true When using Typedef, the types are still the same: alias T1New = Typedef!(T1); alias T2New = Typedef!(T2); writeln(__traits(isSame, T1New, T2New)); // still prints true The documentation of Typedef says: "Typedef allows the creation of a unique type which is based on an existing type. Unlike the alias feature, Typedef ensures the two types are not considered as equals." Shouldn't the second part at least print false then? as for `Typedef!` -- you can use it's third arg, "cookie": import std.typecons; alias T1 = Tuple!(int, int); alias T2 = Tuple!(int, int); alias T1New = Typedef!(T1, T1.init, "t0"); alias T2New = Typedef!(T2, T2.init, "t1"); pragma(msg, __traits(isSame, T1New, T2New)); // false there was a heated discussion about `std.typecons.Typedef`, built-in `typedef` and other related things, but the decision was to keep the status quo. I can't believe I missed the cookie part. Thanks!
Re: Tuple/Typedef question
On Sun, 11 Jan 2015 11:41:08 + Martin via Digitalmars-d-learn wrote: > Is there a way to get Tuple (and Typedef) from the std.typecons > module to generate a new type that is unique on every > instantiation? What I mean is: > > alias T1 = Tuple!(int, int); > alias T2 = Tuple!(int, int); > > writeln(__traits(isSame, T1, T2)); // prints true > > When using Typedef, the types are still the same: > > alias T1New = Typedef!(T1); > alias T2New = Typedef!(T2); > > writeln(__traits(isSame, T1New, T2New)); // still prints true > > The documentation of Typedef says: > "Typedef allows the creation of a unique type which is based on > an existing type. Unlike the alias feature, Typedef ensures the > two types are not considered as equals." > > Shouldn't the second part at least print false then? as for `Typedef!` -- you can use it's third arg, "cookie": import std.typecons; alias T1 = Tuple!(int, int); alias T2 = Tuple!(int, int); alias T1New = Typedef!(T1, T1.init, "t0"); alias T2New = Typedef!(T2, T2.init, "t1"); pragma(msg, __traits(isSame, T1New, T2New)); // false there was a heated discussion about `std.typecons.Typedef`, built-in `typedef` and other related things, but the decision was to keep the status quo. signature.asc Description: PGP signature
Tuple/Typedef question
Is there a way to get Tuple (and Typedef) from the std.typecons module to generate a new type that is unique on every instantiation? What I mean is: alias T1 = Tuple!(int, int); alias T2 = Tuple!(int, int); writeln(__traits(isSame, T1, T2)); // prints true When using Typedef, the types are still the same: alias T1New = Typedef!(T1); alias T2New = Typedef!(T2); writeln(__traits(isSame, T1New, T2New)); // still prints true The documentation of Typedef says: "Typedef allows the creation of a unique type which is based on an existing type. Unlike the alias feature, Typedef ensures the two types are not considered as equals." Shouldn't the second part at least print false then?