Re: Map with maintained insertion order

[Michael Rynn]

On Fri, 23 Mar 2012 23:48:51 +0100, Andrej Mitrovic wrote:

> Does someone have a map implementation that maintains the insertion
> order of the keys?
> 
> E.g.:
> 
> string[string] map;
> map["foo"] = "x";
> map["bar"] = "y";
> 
> When iterating over map keys I want to visit "foo" first, then "bar",
> and so on.

http://bazaar.launchpad.net/~michael-rynn-500/d2-xml/trunk/view/head:/alt/
arraymap.d

I took and adapted the idea from Ultimate++ library code. Ultimate++ is 
worth a try, even if it is C++.

arraymap.d implements a hashmap, that will maintain insertion order, but 
only if no deletions are made.

Each added key-value pair is appended into a separate key and values 
array.

Also maintained is an array of hash_t.

Also maintained is an array of index integers. Matching hash keys are 
related by the linked list method (no pointers).

So starting with an empty map, each insertion appends the key and value 
array storage.   But a key removal will create a hole, and the hole is 
added to the free links list.  The hash_t array points to one of the 
linked list entries which indexes the key/value array.

arraymap.d would appear to have the property, that the hash values cannot 
be mistaken as aliased pointers by the GC, because they can be stored in 
a non-scanned array.

In the same module is a template version copycat implementation of 
druntime AA, aahash.d  with some customizations that I have played around 
with, like being able to use char[] for lookups to string keys (but not 
for op_put). aahash module requires hashutil.d and blockheap.  arraymap 
requires hashutil.d


In performance tests the arraymap.d was slower than the D runtime look 
alike, but acceptable.  I think it has scope for improvement in coding 
and performance.

Rewrite of std.range docs (Was: Re: Making sense of ranges)

[H. S. Teoh]

On Sat, Mar 24, 2012 at 06:19:32PM +, Stewart Gordon wrote:
> The documentation for std.range states
> 
> http://dlang.org/phobos/std_range.html
> "This module defines the notion of range (by the membership tests
> isInputRange, isForwardRange, isBidirectionalRange,
> isRandomAccessRange), range capability tests (such as hasLength or
> hasSlicing), and a few useful range incarnations."
> 
> But that intro doesn't describe what a range actually is.
[...]

In a previous post long lost inside a rather large thread some time ago,
I proposed a rewrite of the docs of std.range to make it clearer, at
least on a basic level, what a range is, why we should care, and what
the module offers.

This thread has further convinced me that std.range's docs *need* this
rewrite. So here's my first attempt at it:

https://github.com/quickfur/phobos/tree/stdrange_docs

It's not complete yet; I still have a few more range-creation templates
to cover, and then list the auxiliary functions, etc., before it's ready
for merging into Phobos. But I thought I should put it up for review
here first, in case people have some comments/suggestions or further
things that should be put into the docs. Better to put it all in a
single pull request than waste the Phobos maintainers' time with
multiple updates to the docs.

So, comments, suggestions, flames, etc., are hereby solicited. :-)


T

-- 
If lightning were to ever strike an orchestra, it'd always hit the conductor 
first.

Re: opDispatch(string name, E...) (E e) question.

[Artur Skawina]

On 03/26/12 00:58, bls wrote:
> Thanks Artur,
> 
> On 03/25/2012 03:18 PM, Artur Skawina wrote:
>> On 03/25/12 22:45, bls wrote:
>>> How do I "call" opDispatch(string name, E...)(E elements) ?
>>> What I want to archive is to call f.i. fm.list with an arbitrary number of 
>>> arguments without using
>>>
>>> fm.list(1, "abc", 4L, 3.33);
>>>
>>> Instead I would prefer
>>> fm.list = (1, "abc", 4L, 3.33);
>>>
>>> Is this somehow possible ?
>>
>> Well, you can do
>>
>> template ID(A...) { alias A ID; }
>>
>> fm.list = ID!(1, "abc", 4L, 3.33);
>>
>> but is that any better than your first version above?...
> Not sure...
> 
> Maybe if we rename it
> fm.list = Values!(1,2,true);
> 
> But I does not work..  Seems I am not able to figure out the opDispatch 
> signature.

import std.stdio;

template Values(A...) { alias A Values; }

struct FlexMap {
   template opDispatch(string key) {
  auto opDispatch(VS...)(VS vs) {
 foreach (v; vs)
writefln("%s: (%s)%s;", key, typeof(v).stringof, v);
  }
   }
}

void main() {
   auto fm = FlexMap();
   fm.list = Values!(1, "abc", 4L, 3.33);
}

artur

Re: opDispatch(string name, E...) (E e) question.

[Simen Kjærås]

On Sun, 25 Mar 2012 22:45:57 +0200, bls  wrote:


How do I "call" opDispatch(string name, E...)(E elements) ?
What I want to archive is to call f.i. fm.list with an arbitrary number  
of arguments without using


fm.list(1, "abc", 4L, 3.33);

Instead I would prefer
fm.list = (1, "abc", 4L, 3.33);

Is this somehow possible ?


No. However, this is:

fm.list = tuple( 1, "abc", 4L, 3.33 );

In that case, you probably want to use the overload
opDispatch(string name, E...)(Tuple!E elements)

and mark the other:
opDispatch(string name, E...)(E element) if ( !(E.length == 1 &&  
isTuple!(E[0])))


tuple, Tuple, and isTuple are found in std.typecons.



import std.variant;
import std.conv;

auto fm = FlexMap();

fm.ten = 10;
fm.ten = ["Ten", "Zehn", "Tein"];
fm.list = [20, 10, 2, 2, 44 ] ;
fm.list = "Hello opDispatch";

struct FlexMap
{
 Variant[] [string] map;

 Variant[] opDispatch(string name)()
 {
return map[name];
 }

 Variant[] opDispatch(string name, E...)(E elements)
 {
foreach(element; elements)
map[name] ~= to!Variant(element);
return map[name];
 }

 Variant[] opDispatch(string name, T) (T t)
 {
map[name] ~= to!Variant(t);
return map[name];
 }  
}



Another question :
How do I bring in :

opDispatch(string name, T) (T[] t)

into FlexMap ?



It's possible to do with template constraints:

import std.traits : isArray;

void opDispatch(string name, T)(T t) if ( isArray!T )
{
// Array specific.
}
void opDispatch(string name, T)(T t) if ( !isArray!T)
{
// Single element.
}



One could also use static if:

Variant[] opDispatch(string name, T)(T t)
{
static if ( is( T U : U[] ) )
{
map[name] ~= to!
}
else
{
// Whatever you want to do differently here.
}
}

Re: opDispatch(string name, E...) (E e) question.

[bls]

Thanks Artur,

On 03/25/2012 03:18 PM, Artur Skawina wrote:

On 03/25/12 22:45, bls wrote:

How do I "call" opDispatch(string name, E...)(E elements) ?
What I want to archive is to call f.i. fm.list with an arbitrary number of 
arguments without using

fm.list(1, "abc", 4L, 3.33);

Instead I would prefer
fm.list = (1, "abc", 4L, 3.33);

Is this somehow possible ?


Well, you can do

template ID(A...) { alias A ID; }

fm.list = ID!(1, "abc", 4L, 3.33);

but is that any better than your first version above?...

Not sure...

Maybe if we rename it
fm.list = Values!(1,2,true);

But I does not work..  Seems I am not able to figure out the opDispatch 
signature.


Think I will rewrite it without using opDispatch.



artur

Re: opDispatch(string name, E...) (E e) question.

[Artur Skawina]

On 03/25/12 22:45, bls wrote:
> How do I "call" opDispatch(string name, E...)(E elements) ?
> What I want to archive is to call f.i. fm.list with an arbitrary number of 
> arguments without using
> 
> fm.list(1, "abc", 4L, 3.33);
> 
> Instead I would prefer
> fm.list = (1, "abc", 4L, 3.33);
> 
> Is this somehow possible ?

Well, you can do

   template ID(A...) { alias A ID; }

   fm.list = ID!(1, "abc", 4L, 3.33);

but is that any better than your first version above?...

artur

Re: opDispatch(string name, E...) (E e) question.

[bls]

On 03/25/2012 02:59 PM, James Miller wrote:

Ok, so looking here:http://dlang.org/function.html, I have determined
that, if you are using Variant arrays (though I'm not sure if you can
do that using literals...) you can use the syntax from this example:


Thanks James..
will give it tomorrow a new try.

At least
Variant[] va =  [1, 2.3222, "abc"];

is not working.

Guess I have to give up the opDispatch() thing and create it a bit more 
traditional :)



Bjoern

oh, beside you mean :
opDispatch(string name, T) (T[] t...)

Re: opDispatch(string name, E...) (E e) question.

[James Miller]

On 26 March 2012 10:34, bls  wrote:
>
> (T) (T[] t) AND (T) (T t) seems not to work.

Ok, so looking here: http://dlang.org/function.html, I have determined
that, if you are using Variant arrays (though I'm not sure if you can
do that using literals...) you can use the syntax from this example:

int test() {
 return sum(1, 2, 3) + sum(); // returns 6+0
}

int func() {
 int[3] ii = [4, 5, 6];
 return sum(ii); // returns 15
}

int sum(int[] ar ...) {
 int s;
 foreach (int x; ar)
  s += x;
 return s;
}

You'll probably need to do some experimentation to figure out how
Variant fits into that properly, but it shouldn't be too hard.

Also, remember that opDispatch takes the name of the function as the
last parameter, so watch out for that.

--
James Miller

Re: opDispatch(string name, E...) (E e) question.

[bls]

On 03/25/2012 02:04 PM, James Miller wrote:

On 26 March 2012 09:45, bls  wrote:

How do I "call" opDispatch(string name, E...)(E elements) ?
What I want to archive is to call f.i. fm.list with an arbitrary number of
arguments without using

fm.list(1, "abc", 4L, 3.33);

Instead I would prefer
fm.list = (1, "abc", 4L, 3.33);


You can use @property on opDispatch to use setter/getter notation,
however I wouldn't rely on that functionality long-term if you want to
keep the same function-call syntax (since -property flag is supposed
to enforce proper parenthesis use on `@property`s).


fm.list = (1, "abc", 4L, 3.33);


I'm hoping you mean `fm.list = [1, "abc", 4L, 3.33];` I think that
using the right template parameters, you can use the same code for
(T...)(T el) and (T)(T[]), I just can't remember what that is...



Ouch, yep, I mean [1, "abc", 4L, 3.33]
But I have no clue how to implement it.



Another question :
How do I bring in :

opDispatch(string name, T) (T[] t)


--
James Miller


(T) (T[] t) AND (T) (T t) seems not to work.
snip
struct FlexMap
{
Variant[] [string] map;

Variant[] opDispatch(string name)()
{
   return map[name];
}

Variant[] opDispatch(string name, E...)(E elements)
{
foreach(element; elements)
map[name] ~= to!Variant(element);

return properties[name];
}

Variant[] opDispatch(string name, T) (T t)
{
map[name] ~= to!Variant(t);
return map[name];
}   
// No go
Variant[] opDispatch(string name, T) (T[] t) {}

}

Re: opDispatch(string name, E...) (E e) question.

[James Miller]

On 26 March 2012 09:45, bls  wrote:
> How do I "call" opDispatch(string name, E...)(E elements) ?
> What I want to archive is to call f.i. fm.list with an arbitrary number of
> arguments without using
>
> fm.list(1, "abc", 4L, 3.33);
>
> Instead I would prefer
> fm.list = (1, "abc", 4L, 3.33);

You can use @property on opDispatch to use setter/getter notation,
however I wouldn't rely on that functionality long-term if you want to
keep the same function-call syntax (since -property flag is supposed
to enforce proper parenthesis use on `@property`s).

> fm.list = (1, "abc", 4L, 3.33);

I'm hoping you mean `fm.list = [1, "abc", 4L, 3.33];` I think that
using the right template parameters, you can use the same code for
(T...)(T el) and (T)(T[]), I just can't remember what that is...

> Another question :
> How do I bring in :
>
> opDispatch(string name, T) (T[] t)

--
James Miller

opDispatch(string name, E...) (E e) question.

[bls]

How do I "call" opDispatch(string name, E...)(E elements) ?
What I want to archive is to call f.i. fm.list with an arbitrary number 
of arguments without using


fm.list(1, "abc", 4L, 3.33);

Instead I would prefer
fm.list = (1, "abc", 4L, 3.33);

Is this somehow possible ?

import std.variant;
import std.conv;

auto fm = FlexMap();

fm.ten = 10;
fm.ten = ["Ten", "Zehn", "Tein"];
fm.list = [20, 10, 2, 2, 44 ] ;
fm.list = "Hello opDispatch";

struct FlexMap
{
Variant[] [string] map;

Variant[] opDispatch(string name)()
{
   return map[name];
}

Variant[] opDispatch(string name, E...)(E elements)
{
foreach(element; elements)
map[name] ~= to!Variant(element);
return map[name];
}

Variant[] opDispatch(string name, T) (T t)
{
map[name] ~= to!Variant(t);
return map[name];
}   
}



Another question :
How do I bring in :

opDispatch(string name, T) (T[] t)

into FlexMap ?

TIA, Bjoern

Re: GC collecting "too much"..

[bearophile]

On Sunday, 25 March 2012 at 19:15:05 UTC, simendsjo wrote:
I'm doing some coding against a c library, and Ds GC keeps 
collecting c owned objects (I think - disabling the GC makes 
everything work)


Three alternative solutions:
- Allocate from the C heap the memory that C will need to use, 
and free it manually or in a struct destructor (RAII) or with 
scope(exit).

- Keep a pointer to the D-GC memory in the D code too.
- In core.memory there are ways to disable scanning of a memory 
zone. Maybe it's usable for your purposes too.


Bye,
bearophile

GC collecting "too much"..

[simendsjo]

I'm doing some coding against a c library, and Ds GC keeps collecting c  
owned objects (I think - disabling the GC makes everything work)

But how can I figure out what the GC is (read: I am) fucking up?
I have some to!string(c_struct_field) and format("%s", c_struct_field) and  
field = c_fields[i] etc etc

Re: std.json

[AaronP]

On 03/25/2012 12:50 PM, Andrea Fontana wrote:

Hope it's clear...


import std.json;
import std.stdio;

void main(string args[])
{
JSONValue json = parseJSON(q"EOS
{
"key" :
{
"subkey1" : "str_val",
"subkey2" : [1,2,3],
"subkey3" : 3.1415
}
}
EOS");
writeln(json.object["key"].object["subkey1"].str);
writeln(json.object["key"].object["subkey2"].array[1].integer);
writeln(json.object["key"].object["subkey3"].type == JSON_TYPE.FLOAT);
}


On Sunday, 25 March 2012 at 15:26:31 UTC, AaronP wrote:

Could I get a "hello, world" example of parsing json? The docs look
simple enough, but I could still use an example.




Ah. That's perfect, thanks. :)

Re: std.json

[Andrea Fontana]

Hope it's clear...


import std.json;
import std.stdio;

void main(string args[])
{
JSONValue json = parseJSON(q"EOS
{
"key" :
{
"subkey1" : "str_val",
"subkey2" : [1,2,3],
"subkey3" : 3.1415
}
}
EOS");
writeln(json.object["key"].object["subkey1"].str);

writeln(json.object["key"].object["subkey2"].array[1].integer);
writeln(json.object["key"].object["subkey3"].type == 
JSON_TYPE.FLOAT);

}


On Sunday, 25 March 2012 at 15:26:31 UTC, AaronP wrote:
Could I get a "hello, world" example of parsing json? The docs 
look simple enough, but I could still use an example.

Problem with receiveOnly and classes

[Ghislain]

Hello,

I need to pass objects of a hierarchy between threads and I have some 
troubles.

The sample code below displays:
Unknown
B.fun()

I do not understand why an object of type A is fetched as a Variant, while 
a object of type B is received correctly.

Instead of 
(A a){a.fun();}
I also tried 
(immutable(A) a){a.fun();}
which resulted in 
core.exception.AssertError@/usr/include/d/dmd/phobos/std/variant.d(286): 
immutable(A)

Any idea?
I am using dmd 2.057 with linux.

Thanks

import std.stdio;
import std.concurrency;

class A{
public:
void fun()const{
writeln("A.fun()");
}
};
class B : A{
public:
override void fun()const{
writeln("B.fun()");
}
};

void producer(Tid t){   
auto a = new immutable(A);
auto b = new immutable(B);
t.send(a);
t.send(b);
}

void main(){
auto t = spawn(&producer, thisTid);
while(1){
receive( 
(A a){a.fun();},
(Variant v){writeln("Unknown");}
);
}
}

Re: GUI library

[Jacob Carlborg]

On 2012-03-25 17:22, Kevin Cox wrote:

I would reccomend Qt as well.  You will get native cross-platform
widgets with great performance.  I am not sure how far QtD is but I know
it once had a lot of development on it.


I don't think Qt is uses the native drawing operations on Mac OS X.

--
/Jacob Carlborg

std.json

[AaronP]

Could I get a "hello, world" example of parsing json? The docs look 
simple enough, but I could still use an example.

Re: GUI library

[Kevin Cox]

On Sun, Mar 25, 2012 at 11:13 AM, Jacob Carlborg  wrote:

> On 2012-03-25 15:04, Tyro[17] wrote:
>
>> Is there one available for use with D2 on MAC OS X?
>>
>> Thanks,
>> Andrew
>>
>
> * QtD - Bindings to Qt. Use the native drawing operations of the operating
> system (I think). Available on all platforms. Not sure if this is developed
> any more.
>

I would reccomend Qt as well.  You will get native cross-platform widgets
with great performance.  I am not sure how far QtD is but I know it once
had a lot of development on it.

Re: GUI library

[Jacob Carlborg]

On 2012-03-25 15:04, Tyro[17] wrote:

Is there one available for use with D2 on MAC OS X?

Thanks,
Andrew


I think these are the choices on Mac OS X:

* gtkD - Bindings to GTK. Does not use the native drawing operations of 
the operating system. Available on all platforms.


http://dsource.org/projects/gtkd

* QtD - Bindings to Qt. Use the native drawing operations of the 
operating system (I think). Available on all platforms. Not sure if this 
is developed any more.


http://dsource.org/projects/qtd

* wxD - Bindings to wxWidgets. Use the native drawing operations of the 
operating system. Available on all platforms. Not sure of the status.


http://wxd.sourceforge.net/

It would also be possible to use Cocoa, as you do with Objective-C, but 
that wouldn't be very practically. There's also a DMD fork that directly 
supports interfacing with Objective-C:


http://michelf.com/projects/d-objc/

--
/Jacob Carlborg

GUI library

[Tyro[17]]

Is there one available for use with D2 on MAC OS X?

Thanks,
Andrew

Parser (not parser generators) and ctfe

[Andrea Fontana]

Does it exists a working ctfe parser for xml or ini or something 
suitable for config file?


Something that import() file and reads values at compile times.

For example:
On my (php) website I use an optimized c++ backend (compiled as 
php module). Profiling code I see that website wastes a lot of 
time on opening, reading and parsing config files (a couple of 
files where i read hosts, ports, etc..). If these value was 
compiled as consts it would be a great boost...

Re: trouble calling function from windows dll

[maarten van damme]

While I was playing around some more I noticed that the optimize flag
causes my program to give an access violation while normal compilation
doesn't give any error whatsoever. is this an old bug or have I stumbled
upon a new bug?

Re: trouble calling function from windows dll

[maarten van damme]

Turns out it indeed got the right function pointer and that function is
getting called correctly. What I was trying to do however was forking a
process within another.
One of the things I needed to do was unmapping the base module from memory
from one of the exe's, align correctly and then write the second executable
in the memory of the first one.
NtUnmapViewOfSection seems to fail with error code STATUS_NOT_MAPPED_VIEW.

I've found out this error appears on certain kinds of executable and one's
written in D seem to be one of those. Is there any reason windows complains
that the main module containing the code section is not mapped?

Re: trouble calling function from windows dll

[maarten van damme]

I did not import  ntoskrnl.lib because I'm trying to do everything in user
mode and there I have access to ntdll.dll which contains
ntunmapviewofsection. Thats why I started using implib to create an
ntdll.dll import library but I couldn't get it to work.

It's good to know that it actually returns the right function pointer. Now
I only have to find where the real problem is :)
Thank you for your help.

Re: trouble calling function from windows dll

[John Chapman]

On Saturday, 24 March 2012 at 19:11:38 UTC, maarten van damme 
wrote:

hi,
I'm trying to call NtUnmapViewOfSection from ntdll.dll. 
According to the

msdn docs it should look like

NTSTATUS NtUnmapViewOfSection(
  __in  HANDLE ProcessHandle,
  __in_opt  PVOID BaseAddress
);


I tried to call it by simply declaring

extern(Windows) uint NtUnmapViewOfSection(HANDLE hProcess,PVOID 
baseAddress);


But now I get

 Error 42: Symbol Undefined _NtUnmapViewOfSection@8


I've also tried using GetProcAddress

cast(uint function(HANDLE hProcess,PVOID
address))GetProcAddress(Runtime.loadLibrary("ntdll.dll"),
"NtUnmapViewOfSection")

but when I looked at GetLastError I get error 127 (specified 
procedure

could not be found) and the function doesn't work.


It's likely I'm missing something easy here, I just can't 
figure out what it is.


Someone knows what it is?


Actually, Runtime.loadLibrary should return the function pointer 
correctly.


Your call to GetLastError() is returning 127 because 
Runtime.loadLibrary itself calls GetProcAddress to see if it 
contains a GC-related function (which will fail if it's not a D 
DLL).





Maarten

Re: trouble calling function from windows dll

[John Chapman]

On Saturday, 24 March 2012 at 19:11:38 UTC, maarten van damme 
wrote:

hi,
I'm trying to call NtUnmapViewOfSection from ntdll.dll. 
According to the

msdn docs it should look like

NTSTATUS NtUnmapViewOfSection(
  __in  HANDLE ProcessHandle,
  __in_opt  PVOID BaseAddress
);


I tried to call it by simply declaring

extern(Windows) uint NtUnmapViewOfSection(HANDLE hProcess,PVOID 
baseAddress);


But now I get

 Error 42: Symbol Undefined _NtUnmapViewOfSection@8


Did you import ntoskrnl.lib?




I've also tried using GetProcAddress

cast(uint function(HANDLE hProcess,PVOID
address))GetProcAddress(Runtime.loadLibrary("ntdll.dll"),
"NtUnmapViewOfSection")

but when I looked at GetLastError I get error 127 (specified 
procedure

could not be found) and the function doesn't work.


It's likely I'm missing something easy here, I just can't 
figure out what it is.


Someone knows what it is?


Runtime.loadLibrary is the problem. Use the Win32 LoadLibrary 
instead.





Maarten