Re: Physics libraries in D
On Wednesday, 8 May 2013 at 09:43:43 UTC, Mike Parker wrote: Last I checked, Bullet doesn't have a C interface. That means it's unlikely you'll find a D binding. And I haven't heard of any. If it has a c++ interface probably you can bind it using swig.
Re: Scoped import bug?
On Thursday, 9 May 2013 at 05:01:15 UTC, Mike Linford wrote:
I'm not sure whether or not I've encountered a bug or whether
my understanding of scoped imports is just faulty.
blah.d:
1 module blah;
2
3 version(A)
4 {
5import std.range;
6 }
7
8 struct Blah(R)
9 {
10version(B)
11{
12 import std.range;
13}
14static assert(isInputRange!R);
15
16void blah(R r)
17{
18 version(C)
19 {
20 assert(r.front == 'h');
21 }
22}
23 }
24
25 void main()
26 {
27Blah!string blah;
28blah.blah(hello);
29 }
Results:
rdmd -version=A -version=C blah
Compiles fine. Module-level import works for the static assert
on line 14 and the runtime assert on line 20
rdmd -version=B blah
Compiles fine! Struct-level import works for the static assert
on line 14
rdmd -version=B -version=C blah
Fails to compile. I get the following error messages:
blah.d(20): Error: no property 'front' for type 'string'
blah.d(27): Error: template instance blah.Blah!(string) error
instantiating
So it appears that when my import is at the struct level like
line 12 I'm able to use the static assert but *not* the runtime
assert. Why cant the function Blah.blah() find the array
implementation of front?
This is a known UFCS name look-up issue.
In 2.063dev, the bug is fixed. Please wait the release.
Kenji Hara
Template alias parameter does not accept types
Version D 2.062
http://dlang.org/template.html#TemplateAliasParameter
Is is said in the documentation that is's possible but i get
compile time error.
template GetString(alias Arg)
{
enum string GetString = Arg.stringof;
}
void main(string[] argv)
{
writeln(GetString!1234);
writeln(GetString!18);
writeln(GetString!int); // Error: template instance
GetString!(int)
// GetString!(int) does not match template
// declaration GetString(alias Arg)
readln();
}
Re: Template alias parameter does not accept types
ref2401:
template GetString(alias Arg)
{
enum string GetString = Arg.stringof;
}
...
writeln(GetString!int); // Error: template instance
Template alias arguments don't yet accept built-in types as int.
It will be fixed.
Bye,
bearophile
Tricky code with exceptions
A little Java program I've just found in a blog post:
class Flow {
static public void main(String[] args) {
for (int i = 0; i 6; ++i) {
System.out.println(Loop: + i);
try {
try {
if (i == 3)
break;
} finally {
if (i % 2 != 0)
throw new Exception();
}
} catch (Exception e) {
System.out.println(Caught);
}
}
}
}
Its output:
Loop: 0
Loop: 1
Caught
Loop: 2
Loop: 3
Caught
Loop: 4
Loop: 5
Caught
My D translation:
import std.stdio;
void main() {
foreach (i; 0 .. 6) {
writeln(Loop: , i);
try {
try {
if (i == 3)
break;
} finally {
if (i % 2 != 0)
throw new Exception();
}
} catch (Exception e) {
writeln(Caught);
}
}
}
It prints:
Loop: 0
Loop: 1
Caught
Loop: 2
Loop: 3
And then it crashes.
Bye,
bearophile
Re: Template alias parameter does not accept types
On Thursday, 9 May 2013 at 10:59:02 UTC, ref2401 wrote:
Version D 2.062
http://dlang.org/template.html#TemplateAliasParameter
Is is said in the documentation that is's possible but i get
compile time error.
template GetString(alias Arg)
{
enum string GetString = Arg.stringof;
}
void main(string[] argv)
{
writeln(GetString!1234);
writeln(GetString!18);
writeln(GetString!int); // Error: template instance
GetString!(int)
// GetString!(int) does not match template
// declaration GetString(alias Arg)
readln();
}
You just can't pass by alias built-in types like int or char
Workaround:
import std.stdio;
template GetString(Arg...)
if(Arg.length == 1)
{
enum string GetString = Arg[0].stringof;
}
void main(string[] argv)
{
writeln(GetString!1234);
writeln(GetString!18);
writeln(GetString!int); // Error: template instance
GetString!(int)
// GetString!(int) does not match template
// declaration GetString(alias Arg)
readln();
}
Re: Template alias parameter does not accept types
On Thursday, 9 May 2013 at 11:19:38 UTC, bearophile wrote: It will be fixed. Ugh, proof-link? I have always thought it is by design and that actually makes sense.
Re: Argument S to typeof is not an expression
On Thursday, 9 May 2013 at 01:08:28 UTC, Meta wrote:
I can't get the structs to be evaluated at compile time, so I
guess that's a lost cause.
Well, structs can be evaluated at compile-time but can't be value
template parameters. You can do a small trick in this case though:
http://dpaste.1azy.net/c2188c90
struct Test(int N)
{
enum n = N;
auto add(int i)()
{
return Test!(n + i)();
}
alias n this;
}
void main()
{
import std.stdio;
enum a = Test!5();
enum b = Test!3();
writeln(a.add!b());
}
Will this do?
But actually this feels more like a task for CTFE. You can write
a normal opAdd that does bounds check at run-time and then add
if (!__ctfe) assert(0); to guarantee it will never be used in
real run-time.
Re: Tricky code with exceptions
bearophile bearophileh...@lycos.com wrote in message
news:pnwldlckpgrjvvuje...@forum.dlang.org...
SNIP
My D translation:
import std.stdio;
void main() {
foreach (i; 0 .. 6) {
writeln(Loop: , i);
try {
try {
if (i == 3)
break;
} finally {
if (i % 2 != 0)
throw new Exception();
}
} catch (Exception e) {
writeln(Caught);
}
}
}
It prints:
Loop: 0
Loop: 1
Caught
Loop: 2
Loop: 3
And then it crashes.
Bye,
bearophile
Strangely, if you replace the break instruction with continue (I know
it's pointless code), it also crashes...
Regards, Mike.
Re: Argument S to typeof is not an expression
then add if (!__ctfe) assert(0); to guarantee it will never be used in real run-time. I don't think I've ever seen __ctfe before. Is it specific to DMD?
Re: Tricky code with exceptions
It works on dpaste http://dpaste.dzfl.pl/fcd2f2b5 which seems to be based on linux 2.62. Which platform do you use? P.S. Seems we can define new kind of forum contribution - some code works as expected in language X, but equivalent code in D goes ballistic. This is very sad.
Re: Argument S to typeof is not an expression
On Thursday, 9 May 2013 at 12:35:29 UTC, Meta wrote: then add if (!__ctfe) assert(0); to guarantee it will never be used in real run-time. I don't think I've ever seen __ctfe before. Is it specific to DMD? It is documented here : http://dlang.org/function.html The __ctfe boolean pseudo-variable, which evaluates to true at compile time, but false at run time, can be used to provide an alternative execution path to avoid operations which are forbidden at compile time. Every usage of __ctfe is evaluated before code generation and therefore has no run-time cost, even if no optimizer is used So no, it is quite official and not specific to dmd.
Re: Template alias parameter does not accept types
On Thu, 09 May 2013 06:58:57 -0400, ref2401 refacto...@gmail.com wrote:
Version D 2.062
http://dlang.org/template.html#TemplateAliasParameter
Is is said in the documentation that is's possible but i get compile
time error.
template GetString(alias Arg)
{
enum string GetString = Arg.stringof;
}
void main(string[] argv)
{
writeln(GetString!1234);
writeln(GetString!18);
writeln(GetString!int); // Error: template instance GetString!(int)
// GetString!(int) does not match template
// declaration GetString(alias Arg)
readln();
}
Others have answered, but the technical issue is that alias accepts a
*symbol*, int is a *keyword*.
Annoying, I know, but I don't see how the compiler can make sense of it,
the way it is built. Keywords are special and must go in the right places.
You can alias int, and it works:
alias int intalias;
writeln(GetString!intalias); // writes intalias
You can also overload GetString to accept builtin types:
template GetString(T) if (is(T == int) || is(T == float) || ...)
{
enum string GetString = T.stringof;
}
There may even already be a template for detecting a builtin type, not
sure.
On Thu, 09 May 2013 07:19:37 -0400, bearophile bearophileh...@lycos.com
wrote:
Template alias arguments don't yet accept built-in types as int. It will
be fixed.
I don't think this is true, alias strictly accepts symbols or literals,
not keywords. Do you have evidence of an official statement to the
contrary?
-Steve
Re: Template alias parameter does not accept types
On Thursday, 9 May 2013 at 12:09:03 UTC, Dicebot wrote: On Thursday, 9 May 2013 at 11:19:38 UTC, bearophile wrote: It will be fixed. Ugh, proof-link? I have always thought it is by design and that actually makes sense. AFAIK, there is no plan for fix. The behavior is currently a part of language design. Kenji Hara
Re: Template alias parameter does not accept types
Steven Schveighoffer: Do you have evidence of an official statement to the contrary? Here I don't have evidence, just faith :-) Bye, bearophile
Re: Template alias parameter does not accept types
On Thursday, 9 May 2013 at 13:27:35 UTC, Steven Schveighoffer
wrote:
template GetString(T) if (is(T == int) || is(T == float) || ...)
{
enum string GetString = T.stringof;
}
Current idiomatic D way to handle both built-in types and symbols
looks like this:
template Hello(T...)
if (T.length == 1)
{
static if (is(T[0]))
// ...
else
// ...
}
I actually don't understand why it accepts literals as those are
not symbols and, for example, you can't attach UDA to literal.
Re: Tricky code with exceptions
Maxim Fomin: It works on dpaste http://dpaste.dzfl.pl/fcd2f2b5 which seems to be based on linux 2.62. On dpaste it also works on gdc2.060 and ldc2.060, both 64 bit. Which platform do you use? Vista 32. Probably I will add it to Bugzilla. P.S. Seems we can define new kind of forum contribution - some code works as expected in language X, but equivalent code in D goes ballistic. This is very sad. On the other hand often it's broken code in other languages that works as desired in D :-) Bye, bearophile
Re: Template alias parameter does not accept types
Kenji Hara:
AFAIK, there is no plan for fix. The behavior is currently a
part of language design.
Considering matters of D semantic uniformity, and the code shown
by Dicebot:
template Hello(T...)
if (T.length == 1)
{
static if (is(T[0]))
// ...
else
// ...
}
Then maybe the current behavior is worth reconsidering.
Bye,
bearophile
Re: Template alias parameter does not accept types
On Thu, 09 May 2013 09:38:29 -0400, Dicebot m.stras...@gmail.com wrote:
On Thursday, 9 May 2013 at 13:27:35 UTC, Steven Schveighoffer wrote:
template GetString(T) if (is(T == int) || is(T == float) || ...)
{
enum string GetString = T.stringof;
}
Current idiomatic D way to handle both built-in types and symbols looks
like this:
template Hello(T...)
if (T.length == 1)
{
static if (is(T[0]))
// ...
else
// ...
}
I actually don't understand why it accepts literals as those are not
symbols and, for example, you can't attach UDA to literal.
I think it was a happy accident, or else a relaxation of the rules to
allow more flexibility. Possibly the same reason could be used to relax
the rules again. But I think it would be a more difficult prospect in the
compiler.
alias cannot accept literals in a normal alias statement. But what
allowing it for template alias parameters does is enable the whole
std.algorithm acceptance of a string literal for function lambdas.
Arguably, this is no longer needed due to the new lambda syntax (but I
think we have found other uses for it). It has also caused some headaches:
map!a.x(foo)
is a different instantiation with identical code to:
map! a.x(foo)
-Steve
Re: Template alias parameter does not accept types
Feels like I am the only one who wants to restrict rules, not relax them =/ But, erm, you don't need alias accepting literals for old syntax! Strings are valid as template value parameters: http://dpaste.1azy.net/54e5d3f2
Re: Template alias parameter does not accept types
On Thu, 09 May 2013 10:00:45 -0400, Dicebot m.stras...@gmail.com wrote: Feels like I am the only one who wants to restrict rules, not relax them =/ But, erm, you don't need alias accepting literals for old syntax! Strings are valid as template value parameters: http://dpaste.1azy.net/54e5d3f2 Hm.. maybe it was a bug that turned into a feature :) Perhaps I'm not remembering correctly why this was introduced. It is true that you could simply accept a literal with an int/long or string, and then you can have different template instantiations based on that. -Steve
Re: Tricky code with exceptions
For what it's worth, this runs fine on 64-bit OSX.
Re: Tricky code with exceptions
On Thursday, 9 May 2013 at 11:24:03 UTC, bearophile wrote:
A little Java program I've just found in a blog post:
class Flow {
static public void main(String[] args) {
for (int i = 0; i 6; ++i) {
System.out.println(Loop: + i);
try {
try {
if (i == 3)
break;
} finally {
if (i % 2 != 0)
throw new Exception();
}
} catch (Exception e) {
System.out.println(Caught);
}
}
}
}
Its output:
Loop: 0
Loop: 1
Caught
Loop: 2
Loop: 3
Caught
Loop: 4
Loop: 5
Caught
My D translation:
import std.stdio;
void main() {
foreach (i; 0 .. 6) {
writeln(Loop: , i);
try {
try {
if (i == 3)
break;
} finally {
if (i % 2 != 0)
throw new Exception();
}
} catch (Exception e) {
writeln(Caught);
}
}
}
It prints:
Loop: 0
Loop: 1
Caught
Loop: 2
Loop: 3
And then it crashes.
Bye,
bearophile
I just tested this for you when you hopped in IRC but you left
before I could tell you that a 64-bit Windows dmd build did not
crash and here is the output.
Loop: 0
Loop: 1
Caught
Loop: 2
Loop: 3
Caught
Loop: 4
Loop: 5
Caught
Re: Tricky code with exceptions
On Thursday, 9 May 2013 at 18:24:46 UTC, Brad Anderson wrote: I just tested this for you when you hopped in IRC but you left before I could tell you that a 64-bit Windows dmd build did not crash and here is the output. Oh, and this was dmd 2.062 (just -m64).
Re: Scoped import bug?
On Thursday, 9 May 2013 at 09:52:12 UTC, Kenji Hara wrote:
On Thursday, 9 May 2013 at 05:01:15 UTC, Mike Linford wrote:
I'm not sure whether or not I've encountered a bug or whether
my understanding of scoped imports is just faulty.
blah.d:
1 module blah;
2
3 version(A)
4 {
5import std.range;
6 }
7
8 struct Blah(R)
9 {
10version(B)
11{
12 import std.range;
13}
14static assert(isInputRange!R);
15
16void blah(R r)
17{
18 version(C)
19 {
20 assert(r.front == 'h');
21 }
22}
23 }
24
25 void main()
26 {
27Blah!string blah;
28blah.blah(hello);
29 }
Results:
rdmd -version=A -version=C blah
Compiles fine. Module-level import works for the static assert
on line 14 and the runtime assert on line 20
rdmd -version=B blah
Compiles fine! Struct-level import works for the static assert
on line 14
rdmd -version=B -version=C blah
Fails to compile. I get the following error messages:
blah.d(20): Error: no property 'front' for type 'string'
blah.d(27): Error: template instance blah.Blah!(string) error
instantiating
So it appears that when my import is at the struct level like
line 12 I'm able to use the static assert but *not* the
runtime assert. Why cant the function Blah.blah() find the
array implementation of front?
This is a known UFCS name look-up issue.
In 2.063dev, the bug is fixed. Please wait the release.
Kenji Hara
Thank you for the explanation. Was this the issue?
http://d.puremagic.com/issues/show_bug.cgi?id=6185
Re: Variant associative arrays
On Thu, 09 May 2013 03:29:06 +0200, evilrat wrote:
first doesn't compile with DMD 2.062 as int implicitly not converted to
long.
foo func takes associative array, within this example you can use type
Variant[string] to make life a bit easier(but i can't recommend it for
ur real code cause we don't know the usage).
and for third func here is example code, though it doesn't takes
associative array you can rewrite it to suit ur needs, also notice that
it's a bit dirty and not very readable cause i'm superlazy and quite in
a hurry, so you can improve it yourself
---
void bar(...)
{
import core.vararg;
enum string handler = `.visit! (
// string type handler (string s) = write(s, pos %2 ==
0 ? ':' :
'\n'),
// long type handler (long l) = write(l, pos %2 == 0 ?
':' : '\n')
)();`;
for( int pos = 0; pos _arguments.length ; pos++ )
{
if ( _arguments[pos] == typeid(HDType) )
mixin(`(va_arg!(HDType)(_argptr))` ~ handler );
else if ( _arguments[pos] == typeid(string) )
mixin(`HDType(va_arg!(string)(_argptr))` ~
handler );
else if ( _arguments[pos] == typeid(long) )
mixin(`HDType(va_arg!(long)(_argptr))` ~ handler
);
else
throw new Exception(wrong type);
}
}
I think I had a large const in there when I tested so it worked then.
This gives me a few ideas, but I may try a different API.
Re: Variant associative arrays
On Thu, 09 May 2013 02:33:08 +0200, bearophile wrote: Byron Heads: I have a variant associative array. In the example below I am wondering if there is a way to create the array without having to indicate the variant type on all of the values. Would like to be able to write code like #2, or something cleaner/better for #1. ... // #1 This works foo([first : HDType(John), last : HDType(Doe), phone : HDType(1234546)]); // #2 Wont compile, type mismatch string and long foo([first : John, last : Doe, phone : 1234546]); If your purpose is to write such associative literal without repeating the HDType(), then a possibility is to write the data in a single string, and feed it to a compile-time function that splits it and builds the associative array. Bye, bearophile That approch may work. Will have to experiment with the API more.
kxml help.
Hi,
I decided to try out kxml and I have the following error.
test.d(10): Error: found '.' when expecting ','
test.d(11): Error: found '.' when expecting ','
when the following is compiled.
module test;
import kxml.xml;
private import std.string;
private import std.stdio;
void main()
{
auto node = new XmlNode();
node.addChild(new XmlNode(mynode).setAttribute(x, 50).
addChild(new XmlNode(Waldo).addCData(Hello!)));
}
-
it's taken out of an example in the code.
Zz
Re: kxml help.
On Thursday, 9 May 2013 at 19:32:44 UTC, Zz wrote:
Hi,
I decided to try out kxml and I have the following error.
test.d(10): Error: found '.' when expecting ','
test.d(11): Error: found '.' when expecting ','
when the following is compiled.
module test;
import kxml.xml;
private import std.string;
private import std.stdio;
void main()
{
auto node = new XmlNode();
node.addChild(new XmlNode(mynode).setAttribute(x, 50).
addChild(new XmlNode(Waldo).addCData(Hello!)));
}
-
it's taken out of an example in the code.
Zz
You just need brackets around the new statements:
node.addChild((new XmlNode(mynode)).setAttribute(x, 50).
addChild((new XmlNode(Waldo)).addCData(Hello!)));
Re: Tricky code with exceptions
Brad Anderson: a 64-bit Windows dmd build did not crash and here is the output. Thank you for the data point. So maybe the problem is only on 32 bit Windows. Bye, bearophile
Re: kxml help.
Thanks
On Thursday, 9 May 2013 at 20:26:28 UTC, John Colvin wrote:
On Thursday, 9 May 2013 at 19:32:44 UTC, Zz wrote:
Hi,
I decided to try out kxml and I have the following error.
test.d(10): Error: found '.' when expecting ','
test.d(11): Error: found '.' when expecting ','
when the following is compiled.
module test;
import kxml.xml;
private import std.string;
private import std.stdio;
void main()
{
auto node = new XmlNode();
node.addChild(new XmlNode(mynode).setAttribute(x, 50).
addChild(new XmlNode(Waldo).addCData(Hello!)));
}
-
it's taken out of an example in the code.
Zz
You just need brackets around the new statements:
node.addChild((new XmlNode(mynode)).setAttribute(x, 50).
addChild((new XmlNode(Waldo)).addCData(Hello!)));
Re: Tricky code with exceptions
On Thursday, 9 May 2013 at 20:33:17 UTC, bearophile wrote: Brad Anderson: a 64-bit Windows dmd build did not crash and here is the output. Thank you for the data point. So maybe the problem is only on 32 bit Windows. Bye, bearophile win8 dmd 2.062 32-bit also crashes
Re: Deallocate array?
Hi again Steve, or anyone else whose reading this.
I was playing around with D and C# and need help getting this
array stuff to work.
My D program:
import std.stdio, std.cstream, std.array;
void main()
{
string[] temp_array = new string[1];
for(int i=0;i1;i++)
{
temp_array[i] = a;
}
temp_array = null;
temp_array = new string[1];
for(int i=0;i1;i++)
{
temp_array[i] = a;
}
temp_array = null;
writeln(end);
din.getc();
}
This program causes an OutOfMemoryError.
My C# program:
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
namespace testing
{
class Program
{
static void Main(string[] args)
{
string[] temp_array = new string[1];
for (int i = 0; i 1; i++)
{
temp_array[i] = a;
}
temp_array = null;
temp_array = new string[1];
for (int i = 0; i 1; i++)
{
temp_array[i] = a;
}
temp_array = null;
temp_array = new string[1];
for (int i = 0; i 1; i++)
{
temp_array[i] = a;
}
temp_array = null;
Console.WriteLine(end);
Console.ReadKey(true);
}
}
}
It's the same program, but doesn't cause and OutOfMemory
exception, even if i double the times i run the loop part.
My question is what is how do i get the same C# functionality in
D without running out of memory and preferably without using the
GC directly?
I'm running Windows Vista 64bit, DMD 2.062 fresh install,
compiling with -m32 flag.
Re: Tricky code with exceptions
Tested on Linux - Kubuntu 12.04 64bits dmd 2.058 -m32 dmd 2.058 64 bits dmd 2.062 -m32 dmd 2.062 and the output is this for all of the above: Loop: 0 Loop: 1 Caught Loop: 2 Loop: 3 Caught Loop: 4 Loop: 5 Caught Then, tested on Windows XP SP3 32 bits, dmd 2.062, and it DIDN'T CRASH. The output is: Loop: 0 Loop: 1 Caught Loop: 2 Loop: 3 --jm
Re: Tricky code with exceptions
On Friday, 10 May 2013 at 01:05:39 UTC, Juan Manuel Cabo wrote: Then, tested on Windows XP SP3 32 bits, dmd 2.062, and it DIDN'T CRASH. The output is: Mm, sorry!!, it did crash but I had dr watson disabled in that VM. Confirmed adding a writeln(finished) after the foreach, which isn't reached. So It DID CRASH in Windows XP 32 bits. Sorry for the noise. --jm
Re: Deallocate array?
On 05/09/2013 05:58 PM, Matic Kukovec wrote:
Hi again Steve, or anyone else whose reading this.
I was playing around with D and C# and need help getting this array
stuff to work.
[...]
My question is what is how do i get the same C# functionality in D
without running out of memory and preferably without using the GC
directly?
I'm running Windows Vista 64bit, DMD 2.062 fresh install, compiling with
-m32 flag.
You have a string array of 100_000_000 elements. (You can insert
underscores in literals to make code more readable.)
On a 32-bit system string is 8 bytes (e.g. by pragma(msg,
string.sizeof)). So you have a single array of 800 million bytes. Then,
each of those strings point at 22 bytes of individully allocated memory
areas.
Note that 800 million is about 20% of the entire addressable memory of
that system. If you have a single value that happens to look like a
pointer into that memory, that huge block will remain in memory forever.
The funny thing about dmd's current conservative GC is that even an
integer can be mistaken to be a pointer into that memory.
So that's the problem today.
I don't know the exact requirement here but if you really must have 100
million strings at one time, then you may want to do your own memory
management. The following should work:
import std.stdio;
import core.memory;
void main()
{
enum size_t times = 10;
enum size_t stringCount = 100_000_000;
foreach (i; 0 .. times) {
auto rawMemory = cast(string*)GC.calloc(string.sizeof *
stringCount);
// D's cool feature of making a slice from raw pointer
auto tempArray = rawMemory[0 .. stringCount];
foreach (ref s; tempArray) {
s = a;
}
GC.free(rawMemory);
// You may want to set tempArray to null at this point to prevent
// using it accidentally but it is not necessary.
writefln(Done with allocation %s/%s; please press Enter,
i + 1, times);
readln();
}
}
Ali
Re: Deallocate array?
On Thu, 09 May 2013 22:14:41 -0400, Ali Çehreli acehr...@yahoo.com wrote:
On a 32-bit system string is 8 bytes (e.g. by pragma(msg,
string.sizeof)). So you have a single array of 800 million bytes. Then,
each of those strings point at 22 bytes of individully allocated memory
areas.
No, each array points at static data. Strings are immutables stored in
the data segment.
Note that 800 million is about 20% of the entire addressable memory of
that system. If you have a single value that happens to look like a
pointer into that memory, that huge block will remain in memory forever.
The funny thing about dmd's current conservative GC is that even an
integer can be mistaken to be a pointer into that memory.
So that's the problem today.
Correct, the only solution at the moment is to deallocate explicitly.
It is important to keep in mind that C# has a *precise* garbage collector,
D has a *conservative* garbage collector. C# will always do better than D
at GC as long as this is true.
64 bit address space also helps tremendously.
I don't know the exact requirement here but if you really must have 100
million strings at one time, then you may want to do your own memory
management. The following should work:
import std.stdio;
import core.memory;
void main()
{
enum size_t times = 10;
enum size_t stringCount = 100_000_000;
foreach (i; 0 .. times) {
auto rawMemory = cast(string*)GC.calloc(string.sizeof *
stringCount);
// D's cool feature of making a slice from raw pointer
auto tempArray = rawMemory[0 .. stringCount];
You can allocate normally, and just use free on the pointer
foreach (ref s; tempArray) {
s = a;
}
GC.free(rawMemory);
GC.free(tempArray.ptr); // also works
// You may want to set tempArray to null at this point to
prevent
// using it accidentally but it is not necessary.
writefln(Done with allocation %s/%s; please press Enter,
i + 1, times);
readln();
}
}
-Steve
Re: Deallocate array?
On 05/09/2013 07:43 PM, Steven Schveighoffer wrote: On Thu, 09 May 2013 22:14:41 -0400, Ali Çehreli acehr...@yahoo.com wrote: Then, each of those strings point at 22 bytes of individully allocated memory areas. No, each array points at static data. Strings are immutables stored in the data segment. You are right. They start at pointing to the same string but as soon as each string is used in a special way, say each gets appended a char, then they own their special characters: s = a; // Some time later: s ~= 'x'; Now each s.ptr is different. Ali
Re: Tricky code with exceptions
evilrat evilrat...@gmail.com writes: On Thursday, 9 May 2013 at 20:33:17 UTC, bearophile wrote: Brad Anderson: a 64-bit Windows dmd build did not crash and here is the output. Thank you for the data point. So maybe the problem is only on 32 bit Windows. Bye, bearophile win8 dmd 2.062 32-bit also crashes Anybody looked at the assembly for the crashing executables? The truth is in there.
Re: Scoped import bug?
On Thursday, 9 May 2013 at 18:25:39 UTC, Mike Linford wrote:
On Thursday, 9 May 2013 at 09:52:12 UTC, Kenji Hara wrote:
On Thursday, 9 May 2013 at 05:01:15 UTC, Mike Linford wrote:
I'm not sure whether or not I've encountered a bug or whether
my understanding of scoped imports is just faulty.
blah.d:
1 module blah;
2
3 version(A)
4 {
5import std.range;
6 }
7
8 struct Blah(R)
9 {
10version(B)
11{
12 import std.range;
13}
14static assert(isInputRange!R);
15
16void blah(R r)
17{
18 version(C)
19 {
20 assert(r.front == 'h');
21 }
22}
23 }
24
25 void main()
26 {
27Blah!string blah;
28blah.blah(hello);
29 }
Results:
rdmd -version=A -version=C blah
Compiles fine. Module-level import works for the static
assert on line 14 and the runtime assert on line 20
rdmd -version=B blah
Compiles fine! Struct-level import works for the static
assert on line 14
rdmd -version=B -version=C blah
Fails to compile. I get the following error messages:
blah.d(20): Error: no property 'front' for type 'string'
blah.d(27): Error: template instance blah.Blah!(string) error
instantiating
So it appears that when my import is at the struct level like
line 12 I'm able to use the static assert but *not* the
runtime assert. Why cant the function Blah.blah() find the
array implementation of front?
This is a known UFCS name look-up issue.
In 2.063dev, the bug is fixed. Please wait the release.
Kenji Hara
Thank you for the explanation. Was this the issue?
http://d.puremagic.com/issues/show_bug.cgi?id=6185
Yes, that's right!
Kenji Hara
