digitalmars.D.learn - Move and CTFE
- Q. Schroll (25/25) May 30 2018 It seems one cannot std.algorithm.mutation.move objects
- Jonathan M Davis (15/41) May 30 2018 Why is it suspicious? memcpy is a C function, and you can't call C funct...
- Q. Schroll (31/57) May 30 2018 I definitely do understand the error message and it makes sense
- Jonathan M Davis (18/80) May 30 2018 I'm not sure that it really makes sense to worry about fixing stuff like
- Stefan Koch (16/20) Jun 21 2018 To give some more context here:
- Per =?UTF-8?B?Tm9yZGzDtnc=?= (2/10) Jun 21 2018 Keep up the great work!
It seems one cannot std.algorithm.mutation.move objects
explicitly. Say I have a non-copyable type
struct NoCopy
{
int payload; // some payload
pure nothrow nogc safe disable:
this(this); // make it non copyable
}
that is being used in a compile-time function evaluation where
values are being moved.
int f() pure nothrow nogc safe
{
import std.algorithm.mutation : move;
NoCopy nc = NoCopy(1), nc2 = NoCopy(3);
nc = move(nc2);
return 0;
}
static assert(f() == 0); // trigger CTFE
It fails because move() cannot be executed at compile time. The
reason
"memcpy cannot be interpreted at compile time, because it has
no available source code"
sounds very suspicious. Shouldn't it be possible to move at CTFE,
too, or does it mean, non-copyable types are practically unusable
for CTFE?
May 30 2018
On Wednesday, May 30, 2018 20:42:38 Q. Schroll via Digitalmars-d-learn wrote:It seems one cannot std.algorithm.mutation.move objects explicitly. Say I have a non-copyable type struct NoCopy { int payload; // some payload pure nothrow nogc safe disable: this(this); // make it non copyable } that is being used in a compile-time function evaluation where values are being moved. int f() pure nothrow nogc safe { import std.algorithm.mutation : move; NoCopy nc = NoCopy(1), nc2 = NoCopy(3); nc = move(nc2); return 0; } static assert(f() == 0); // trigger CTFE It fails because move() cannot be executed at compile time. The reason "memcpy cannot be interpreted at compile time, because it has no available source code" sounds very suspicious.Why is it suspicious? memcpy is a C function, and you can't call C functions during CTFE precisely because the compiler doesn't have their source code. You can't call D functions either if the compiler doesn't have their source (e.g. if you're using a .di file to hide the implementation).Shouldn't it be possible to move at CTFE, too, or does it mean, non-copyable types are practically unusable for CTFE?You can't do much in the way of pointer or memory manipulation during CTFE (e.g. no pointer arithmetic or reinterpret casts). So, I don't see how a move could be done during CTFE. Even if the source for memcpy were available during CTFE, I suspect that it wouldn't be allowed due to the lower level stuff that it does. Maybe the newCTFE stuff that Stefan is working on can do more in this area (I don't know), but in general, anything that's at all low-level is forbidden in CTFE. - Jonathan M Davis
May 30 2018
On Wednesday, 30 May 2018 at 21:02:07 UTC, Jonathan M Davis wrote:On Wednesday, May 30, 2018 20:42:38 Q. Schroll via Digitalmars-d-learn wrote:I definitely do understand the error message and it makes sense that it fails the way it's implemented. However, it makes no sense that moving as a concept can fail at CTFE. That's what I find suspicious. [Maybe 'suspicious' isn't the right term; I couldn't express it better.] You can move rvalues at CTFE which proves that the compiler can do it.It seems one cannot std.algorithm.mutation.move objects explicitly. Say I have a non-copyable type [...] It fails because move() cannot be executed at compile time. The reason "memcpy cannot be interpreted at compile time, because it has no available source code" sounds very suspicious.Why is it suspicious? memcpy is a C function, and you can't call C functions during CTFE precisely because the compiler doesn't have their source code. You can't call D functions either if the compiler doesn't have their source (e.g. if you're using a .di file to hide the implementation).That's the explanation why probably all currently possible library alternatives to memcpy would fail. I suspected that when encountering the error, but still wonder why memcpy or other low-level stuff is even necessary to accomplish something the compiler is perfectly able to do. From what I see, the reason for the hack is lack of expressiveness: We don't have rvalue-refs in D (which I find good) so, currently, there is no cast-solution as in C++. So for a proper move() that works at CTFE, we'd need some specific tool. I have no idea of the details on how the compiler handles lvalues. Would it make sense to add a compiler trait, specifically to solve moving? Like __traits(move, lvalue_expression) [name up for discussion] that is identical to lvalue_expression with the exception that the (lvalue/rvalue) flag (or whatever it is) is set to "rvalue". Basically, it's the C++ solution: After the "cast", the compiler will proceed and pretend it is an rvalue and therefore initiate moving. Do you think adding a trait to make move() and swap() work at CTFE is worth it? A quick search showed me the class "Expression" has "virtual bool isLvalue();" so it might be easy as wrapping and hooking that virtual method. To me, [1] highly suggests that it works. [1] https://github.com/dlang/dmd/blob/master/src/dmd/expression.d#L1219Shouldn't it be possible to move at CTFE, too, or does it mean, non-copyable types are practically unusable for CTFE?You can't do much in the way of pointer or memory manipulation during CTFE (e.g. no pointer arithmetic or reinterpret casts). So, I don't see how a move could be done during CTFE. Even if the source for memcpy were available during CTFE, I suspect that it wouldn't be allowed due to the lower level stuff that it does.
May 30 2018
On Wednesday, May 30, 2018 22:42:13 Q. Schroll via Digitalmars-d-learn wrote:On Wednesday, 30 May 2018 at 21:02:07 UTC, Jonathan M Davis wrote:I'm not sure that it really makes sense to worry about fixing stuff like this in CTFE before newCTFE is actually merged. For instance, as I understand it, current CTFE can't even really handle mutation. Rather, it creates a new value every time you mutate a variable. Don explained to me at one point about how even incrementing a variable allocates memory so that you then have a new value to use. Stuff like that is why CTFE is so slow and eats up so much memory. Much as it acts like it's running your code in a normal fashion, it's really not implemented that way (the reason that it works the way does having to do with how it grew into existence out of other features rather than being designed up front). The current CTFE implementation is incredibly hacky, and it's arguably a miracle that it can do as much as it can. newCTFE is taking a very different approach to CTFE, and in theory, it will fix many of the problems that CTFE currently has, but it's taking Stefan quite a while to get it to where it needs to be to actually merge it. - Jonathan M DavisOn Wednesday, May 30, 2018 20:42:38 Q. Schroll via Digitalmars-d-learn wrote:I definitely do understand the error message and it makes sense that it fails the way it's implemented. However, it makes no sense that moving as a concept can fail at CTFE. That's what I find suspicious. [Maybe 'suspicious' isn't the right term; I couldn't express it better.] You can move rvalues at CTFE which proves that the compiler can do it.It seems one cannot std.algorithm.mutation.move objects explicitly. Say I have a non-copyable type [...] It fails because move() cannot be executed at compile time. The reason "memcpy cannot be interpreted at compile time, because it has no available source code" sounds very suspicious.Why is it suspicious? memcpy is a C function, and you can't call C functions during CTFE precisely because the compiler doesn't have their source code. You can't call D functions either if the compiler doesn't have their source (e.g. if you're using a .di file to hide the implementation).That's the explanation why probably all currently possible library alternatives to memcpy would fail. I suspected that when encountering the error, but still wonder why memcpy or other low-level stuff is even necessary to accomplish something the compiler is perfectly able to do. From what I see, the reason for the hack is lack of expressiveness: We don't have rvalue-refs in D (which I find good) so, currently, there is no cast-solution as in C++. So for a proper move() that works at CTFE, we'd need some specific tool. I have no idea of the details on how the compiler handles lvalues. Would it make sense to add a compiler trait, specifically to solve moving? Like __traits(move, lvalue_expression) [name up for discussion] that is identical to lvalue_expression with the exception that the (lvalue/rvalue) flag (or whatever it is) is set to "rvalue". Basically, it's the C++ solution: After the "cast", the compiler will proceed and pretend it is an rvalue and therefore initiate moving. Do you think adding a trait to make move() and swap() work at CTFE is worth it? A quick search showed me the class "Expression" has "virtual bool isLvalue();" so it might be easy as wrapping and hooking that virtual method. To me, [1] highly suggests that it works. [1] https://github.com/dlang/dmd/blob/master/src/dmd/expression.d#L1219Shouldn't it be possible to move at CTFE, too, or does it mean, non-copyable types are practically unusable for CTFE?You can't do much in the way of pointer or memory manipulation during CTFE (e.g. no pointer arithmetic or reinterpret casts). So, I don't see how a move could be done during CTFE. Even if the source for memcpy were available during CTFE, I suspect that it wouldn't be allowed due to the lower level stuff that it does.
May 30 2018
On Wednesday, 30 May 2018 at 23:07:26 UTC, Jonathan M Davis wrote:newCTFE is taking a very different approach to CTFE, and in theory, it will fix many of the problems that CTFE currently has, but it's taking Stefan quite a while to get it to where it needs to be to actually merge it.To give some more context here: I do intend newCTFE to be a drop-in replacement that offers the same functionality (including the very accurate error detection and precise error reporting) Which was quite hard to achieve due to not being able of creating Error-Nodes from inside the interpreter. I'd say for the amount of things that newCTFE currently _correctly_ handles it has been fast progress! CTFE is the most well tested feature in dmd. So there is no room for sloppiness or functional differences! As you previously mentioned the newCTFE engine works on a completely different basis then the old engine does. This does provide both speed and better debugging support, but comes at the cost of having to re-implement a complete backend and some parts of semantic analysis.
Jun 21 2018
Per =?UTF-8?B?Tm9yZGzDtnc=?= <per.nordlow gmail.com> On Thursday, 21 June 2018 at 20:15:42 UTC, Stefan Koch wrote:To give some more context here: CTFE is the most well tested feature in dmd. So there is no room for sloppiness or functional differences! As you previously mentioned the newCTFE engine works on a completely different basis then the old engine does. This does provide both speed and better debugging support, but comes at the cost of having to re-implement a complete backend and some parts of semantic analysis.Keep up the great work!
Jun 21 2018