Is there a way to ignore return values in Ada functions?
I have a function which imports from an Intrinsic.
subtype int32 is Interfaces.Interger_32;
function Intrinsic_Sync_Add_And_Fetch
(P : access int32; I : int32) return int32;
pragma Import(
Intrinsic,
Intrinsic_Sync_Add_And_Fetch,
"__sync_add_and_fetch_4");
If I want to use this in a procedure, I need to accept the return value or I will get a compiler error:
cannot use function Intrinsic_Sync_Add_And_Fetch in procedure call.
But, if I create a variable that simply takes the return value of the function and is never used then I get compiler warnings. Obviously, I'd rather avoid those.
I can't very well assign the value back to the value I'm adding to; this would undermine the point of the add operation being atomic.
There is the option of taking the value and doing something with it, like:
val := Intrinsic_Sync_Add_And_Fetch(...);
if val := 0 then null; end if;
It forces the code to compile without errors or warnings, but it seems stupid to me. How can I "get around" this language feature and safely ignore the return value?
Edit: What is __sync_add_and_fetch_4?
This is a built-in atomic operation available on Intel CPUs. As such, part of my Autoconf/Automake process would be deciding if the operation is available, and use a fallback implementation, which involves a critical section, if it's not.
You can read about this and similar operations in GCC's section on atomic builtins.
The __sync_add_and_fetch_4 does pretty much exactly what it says. In C, it would look something like this:
int32_t __sync_add_and_fetch_4(int32_t *ptr, int32_t value) {
*ptr += value;
return *ptr;
}
So it's an atomic addition operation, which returns the result of the addition. Basically, it's an atomic += operator. The _4 means that it takes a 4-byte integer.
Edit: I understand that I could probably just switch off that particular compiler warning, but that always feels dirty to me. If there's a solution available that allows me to continue using -Wall -Werror then I'd love to see it.
declare
dummy : constant return_type := my_function;
pragma Unreferenced (dummy);
begin null; end;
or write a wrapper procedure.
If you never want to reference the return value, why not declare the subprogram as a procedure? The value is going to be returned in a register, so throwing it away won’t cause a lot of grief. (I stand to be corrected on this one!)
subtype int32 is Interfaces.Integer_32;
procedure Intrinsic_Sync_Add_And_Fetch
(P : access int32; I : int32);
pragma Import(
Intrinsic,
Intrinsic_Sync_Add_And_Fetch,
"__sync_add_and_fetch_4");
You said you're only targeting the GNAT compiler. The GNAT User's Guide says:
Note that a special exemption applies to variables which contain any of the substrings DISCARD, DUMMY, IGNORE, JUNK, UNUSED, in any casing. Such variables are considered likely to be intentionally used in a situation where otherwise a warning would be given, so warnings of this kind are always suppressed for such variables.
So the simplest solution to your problem is :
unused := Intrinsic_Sync_Add_And_Fetch(...);
Though you might want to wrap that in a procedure if you are going to use it more than a couple of times :
procedure Intrinsic_Sync_Add_And_Fetch(P : access int32; I : int32) is
unused : int32;
begin
unused := Intrinsic_Sync_Add_And_Fetch(P : access int32; I : int32);
end Intrinsic_Sync_Add_And_Fetch;
i don't know of any way to ignore the return value of a function in Ada: the language has been especially designed to force you to store those return values.
personally, i would store the return value and ignore any warning regarding the use of the variable. anyway, the said warning is quite strange since the variable is indeed used to store the return value.
Related
I am using the COSMIC compiler in the STVD ide and even though optimization is turned of with -no (documentation says "-no: do not use optimizer") some lines of code get removed and cannot have a breakpoint placed upon them, nor are they to be found in the disassembly.
I tried to set -oc (leave removed instructions as comments) which resulted in not even showing the removed lines as comment.
bool foo(void)
{
uint8_t val;
if (globalvar > 5)
val = 0;
for (val = 0; val < 8; val++)
{
some code...
}
return true;
}
I do know it seems idiotic to set val to 0 prior to the for loop but lets just assume it is for some reason necessary. When I set no optimization I expect it to be not optimized but insted the val = 0; gets removed without any traces.
I am not looking for a workaround like declaring val volatile whitch solves the problem. I am rather looking for a way to prevent the optimization or at least understand/know what changes are made to my code when compiling.
It is not clear from the manual, but it seems that the -no option prevents assembly level optimisation. It seems possible that the code generator stage that runs before assembly optimisation may perform higher level optimisation such as redundant code removal.
From the manual:
-cp
disable the constant propagation optimization. By default,
when a variable is assigned with a constant, any subsequent access to that variable is replaced by the constant
itself until the variable is modified or a flow break is
encountered (function call, loop, label ...).
It seems that it is this constant propagation feature that you must explicitly disable.
It is unusual perhaps, but it appears that this compiler optimises by default, and distinguishes between compiler optimisations and assembler optimisations (performed as the compilation stage), and them makes you switch off each individual optimisation separately.
To avoid this in the code, rather than switching it off globally, you could initialise val to a non-zero value in this case:
int val = -1 ;
Then the later assignment to zero will require explicit code. This has the advantage over volatile perhaps in that it will not block optimisations when you do enable them.
I believe that this behaviour is allowed by the C language specification.
You are effectively writing the same value either once or twice to the same variable on successive lines of code. The compiler could assign this value to either a processor register or a memory location as it sees fit and knows that the value following the initial assignment in the for loop is the same as the value assigned when the if clause is actioned. As a result the language spec allows the compiler to throw the redundant code away.
The way to force the compiler to perform all read and write accesses to the variable is to use the volatile keyword. That is what it is for.
I want to assign literals to some of the variables at the end of the file with my program, but to use these variables earlier. The only method I've come up with to do it is the following:
my $text;
say $text;
BEGIN {
$text = "abc";
}
Is there a better / more idiomatic way?
Just go functional.
Create subroutines instead:
say text();
sub text { "abc" }
UPDATE (Thanks raiph! Incorporating your feedback, including reference to using term:<>):
In the above code, I originally omitted the parentheses for the call to text, but it would be more maintainable to always include them to prevent the parser misunderstanding our intent. For example,
say text(); # "abc"
say text() ~ text(); # "abcabc"
say text; # "abc", interpreted as: say text()
say text ~ text; # ERROR, interpreted as: say text(~text())
sub text { "abc" };
To avoid this, you could make text a term, which effectively makes the bareword text behave the same as text():
say text; # "abc", interpreted as: say text()
say text ~ text; # "abcabc", interpreted as: say text() ~ text()
sub term:<text> { "abc" };
For compile-time optimizations and warnings, we can also add the pure trait to it (thanks Brad Gilbert!). is pure asserts that for a given input, the function "always produces the same output without any additional side effects":
say text; # "abc", interpreted as: say text()
say text ~ text; # "abcabc", interpreted as: say text() ~ text()
sub term:<text> is pure { "abc" };
Unlike Perl 5, in Perl 6 a BEGIN does not have to be a block. However, the lexical definition must be seen before it can be used, so the BEGIN block must be done before the say.
BEGIN my $text = "abc";
say $text;
Not sure whether this constitutes an answer to your question or not.
First, a rephrase of your question:
What options are there for succinctly referring to a variable (or constant etc.) whose initialization code appears further down in the same source file?
Post declare a routine
say foo;
sub foo { 'abc' }
When a P6 compiler parses an identifier that has no sigil, it checks to see if it has already seen a declaration of that identifier. If it hasn't, then it assumes that the identifier corresponds to a routine which will be declared later as a "listop" routine (which takes zero or more arguments) and moves on. (If its assumption turns out to be wrong, it fails the compilation.)
So you can use routines as if they were variables as described in Christopher Bottom's answer.
Autodeclare a variable on first use
strict is a "pragma" that controls how a P6 compiler reacts when it parses an as yet undeclared variable/constant that starts with a sigil.
P6 starts programs with strict mode switched on. This means that the compiler will insist on predeclaration of any sigil'd variable/constant. (By predeclaration I mean an explicit declaration that appears textually before the variable/constant is used.)
But you can write use strict or no strict to control whether the strict pragma is on or off in a given lexical scope, so this will work:
no strict;
say $text;
BEGIN {
$text = "abc";
}
Warning Having no strict in effect (which is unfortunately how most programming languages work) makes accidental misspelling of variable names a bigger nuisance than it is with use strict mode on.
Declare a variable explicitly in the same statement as its first use
You don't have to write a declaration as a separate statement. You can instead declare and use a variable in the same statement or expression:
say my $text;
BEGIN {
$text = "abc";
}
Warning If you repeat my $bar in the exact same lexical scope, the compiler will emit a warning. In contrast, say my $bar = 42; if foo { say my $bar = 99 } creates two distinct $bar variables without warning.
Initialize at run-time
The BEGIN phaser shown above runs at compile-time (after the my declaration, which also happens at compile-time, but before the say, which happens at run-time).
If you want to initialize variables/constants at run-time instead, use INIT instead:
say my $text;
INIT {
$text = "abc";
}
INIT code runs before any other run-time code, so the initialization still happens before the say gets executed.
Use a positronic (ym) variable
Given a literal interpretation of your question a "positronic" or ym variable would be yet another solution. (This feature is not built-in. I'm including it mostly because I encountered it after answering this question and think it belongs here, at the very least for entertainment value.)
Initialization and calculation of such a variable starts in the last statement using it and occurs backwards relative to the textual order of the code.
This is one of the several crazy sounding but actually working and useful concepts that Damian "mad scientist" Conway discusses in his 2011 presentation Temporally Quaquaversal Virtual Nanomachine Programming In Multiple Topologically Connected Quantum-Relativistic Parallel Spacetimes... Made Easy!.
Here's a link to the bit where he focuses on these variables.
(The whole presentation is a delight, especially if you're interested in physics; programming techniques; watching highly creative wunderkinds; and/or enjoy outstanding presentation skills and humor.)
Create a PS pragma?
In terms of coolness, the following pales in comparison to Damian's positronic variable feature that I just covered, but it's an idea I had while pondering this question.
Someone could presumably implement something like the following pragma:
use PS;
say $text;
BEGIN $text = 'abc';
This PS would lexically apply no strict and in addition require that, to avoid a compile-time error:
An auto-declared variable/constant must match up with a post declaration in a BEGIN or INIT phaser;
The declaration must include initialization if the first use (textually) of a variable/constant is not a binding or assignment.
According to the Go reference there are two ways of declaring a variable
Variable_declarations (in the format of var count = 0 or var count int)
and
Short_variable_declarations (in the format of count := 0)
I found it's very confusing to decide which one to use.
The differences I know (till now) are that:
I can only using a count := 0 format when in the scope of a function.
count := 0 can be redeclared in a multi-variable short declaration.
But they do behave the same as far as I know. And in the reference it also says:
It (the count:=0way) is shorthand for a regular variable declaration with initializer expressions but no types
My confusions are:
If one is just the shorthand way of the other, why do they behave differently?
In what concern does the author of Go make two ways of declaring a variable (why are they not merged into one way)? Just to confuse us?
Is there any other aspect that I should keep my eyes open on when using them, in case I fall into a pit?
The Variable declarations make it clear that variables are declared. The var keyword is required, it is short and expresses what is done (at the file level everything excluding comments has to start with a keyword, e.g. package, import, const, type, var, func). Like any other block, variable declarations can be grouped like this:
var (
count int
sum float64
)
You can't do that with Short variable declarations. Also you can use Variable declarations without specifying the initial value in which case each variable will have the zero value of its type. The Short variable declaration does not allow this, you have to specify the initial value.
One of Go's guiding design principle was to make the syntax clean. Many statements require or it is handy that they allow declaring local variables which will be only available in the body of the statement such as for, if, switch etc. To make the syntax cleaner and shorter, Short variable declaration is justified in these cases and it is unambigous what they do.
for idx, value := range array {
// Do something with index and value
}
if num := runtime.NumCPU(); num > 1 {
fmt.Println("Multicore CPU, cores:", num)
}
Another difference: Redeclaration
Quoting from the Language specification:
Unlike regular variable declarations, a short variable declaration may redeclare variables provided they were originally declared earlier in the same block with the same type, and at least one of the non-blank variables is new. As a consequence, redeclaration can only appear in a multi-variable short declaration. Redeclaration does not introduce a new variable; it just assigns a new value to the original.
This one is also handy. Suppose you want to do proper error handling, you can reuse an err variable because most likely you only need it to check if there were any errors during the last function call:
var name = "myfile.txt"
fi, err := os.Stat(name) // fi and err both first declared
if err != nil {
log.Fatal(err)
}
fmt.Println(name, fi.Size(), "bytes")
data, err := ioutil.ReadFile(name) // data is new but err already exists
// so just a new value is assigned to err
if err != nil {
log.Fatal(err)
}
// Do something with data
Let's say I have a Planet:
type Planet is tagged null record;
type Planet_Ref is access Planet'class;
Now I subclass it:
type Habitable_Planet is new Planet with null record;
type Habitable_Planet_Ref is access Habitable_Planet'class;
Now I define some variables:
p: Planet_Ref := Make_Planet;
hp: Habitable_Planet_Ref := Make_Habitable_Planet;
I would naively expect that assigning p := hp would work, because a Habitable_Planet is a subclass of Planet. But of course that won't work because every type defined with type is distinct and doesn't interoperate with any other type.
So I'd expect to have to declare Habitable_Planet_Ref to be a subtype of Planet_Ref to make this work. But the syntax doesn't seem to allow for this.
How do I make this work?
(Yes, I know I can use an explicit view conversion to cast a Habitable_Planet_Ref to a Planet_Ref, but that's really ugly and I'd like to avoid it.)
Ada recognizes types by name, so indeed you would need a view conversion here.
But if you are using Ada 2005, you can use anonymous access types instead. For instance:
hp: access Habitable_Planet'Class := Make_Habitable_Planet;
p: access Planet'Class := hp; -- valid with anonymous access types
One the drawbacks of using anonymous access types is that the code is more
verbose (although in general you would not use them for local variables, but
as parameters to subprograms or as fields in a (tagged) record.
They also can't be used with Unchecked_Deallocation. In fact, I personally often
use them exactly because of that: when I have a field in a record which is of an
anonymous access type, I know that the record does not "own" the accessed data,
and therefore it should not free it (in fact, I would have to write some convoluted
code to free them).
And of course as per your request the result for type matching are slightly more
relax, which is nice too.
ajb is correct in his comment. Ada is too strict for many practices you might be used to in other languages. An alternative would be to just not use objects and instead just simple records or discriminate records. I understand this may not be what you are looking for, but from my experience more can be done with less lines of code and the solution tends to me easier to understand.
Simple record
--...
type Rec_Planet is record
--.. stuff
end record;
--...
type Rec_Habitable_Planet is record
Planet : Rec_Planet := (others => <>);
--.. stuff
end record;
Discriminate record
type Enum_Planet is (Normal_Planet, Habitable_Planet);
type Rec_Planet(Kind : Enum_Planet := Normal_Planet) is record
-- rec_Planet stuff..
case Kind is
when Habitable_Planet => -- Rec_Habitable_Planet stuff
when others => null;
end case;
end record;
So #manuBriot gave me the answer I needed, but there were some other things I was doing wrong in my question which I should clarify because they'll confuse anyone else reading this question.
I was confusing the issue by using accesses. From Ada's point of view all accesses defined with type are distinct, so it never gets as far as looking at what the access is pointing at; it just disallows the assignment.
However, Ada does support implicit upcasting of class-wide types (and also discrete types, where an instance of a subtype will get implicitly cast to its parent type --- implement ALL the class hierarchies! But that's not really relevant here.) Example here:
With Ada.Text_IO; Use Ada.Text_IO;
With Ada.Integer_Text_IO; Use Ada.Integer_Text_IO;
procedure Prog is
package Superclass is
type Class is tagged record
null;
end record;
procedure Announce(self: in out Class);
subtype Var is Class'class;
end;
package body Superclass is
procedure Announce(self: in out Class)
is
begin
Put_Line("I am the superclass");
end;
end;
package Subclass is
type Class is new Superclass.Class with null record;
procedure Announce(self: in out Class);
end;
package body Subclass is
procedure Announce(self: in out Class)
is
begin
Put_Line("I am the subclass");
end;
end;
osuper: Superclass.Class;
osub: Subclass.Class;
vsuper: Superclass.Var := osuper;
vsub: Superclass.Var := osub; -- implicit upclass here
begin
vsuper.Announce;
vsub.Announce;
end;
(It's in ideone here: http://ideone.com/M79l0a Interesting sidenote. If you define subtype Var is Superclass.Var in the Prog package, and then use Var in the definitions of vsuper and vsub, ideone's Ada compiler crashes.)
Of course, like all indefinite types, once the variable has been initialised then its type cannot be changed. So I can assign any Subclass.Object to vsub, but I can't assign a Superclass.Object to it. And of course I'm physically copying the object rather than referring to an object elsewhere.
Implicitly upcasting accesses to class-wide types should be safe. Because assigning to a class-wide type does a runtime instance check to make sure that the physical type of the objects are compatible, it ought not to be possible to accidentally corrupt objects like you can in C++ --- see Overwriting an instance of a subclass with an instance of a superclass, for example. Therefore assigning to a dereferenced access shouldn't be a problem. However, it's 2100 at night and my brain has turned to sludge, so it's entirely possible that I'm missing something here. (Although given that when using anonymous accesses there aren't any problems, I suspect not.) Elucidation welcome...
One of the more notable aspects of Go when coming from C is that the compiler will not build your program if there is an unused variable declared inside of it. So why, then, is this program building if there is an unused parameter declared in a function?
func main() {
print(computron(3, -3));
}
func computron(param_a int, param_b int) int {
return 3 * param_a;
}
There's no official reason, but the reason given on golang-nuts is:
Unused variables are always a programming error, whereas it is common
to write a function that doesn't use all of its arguments.
One could leave those arguments unnamed (using _), but then that might
confuse with functions like
func foo(_ string, _ int) // what's this supposed to do?
The names, even if they're unused, provide important documentation.
Andrew
https://groups.google.com/forum/#!topic/golang-nuts/q09H61oxwWw
Sometimes having unused parameters is important for satisfying interfaces, one example might be a function that operates on a weighted graph. If you want to implement a graph with a uniform cost across all edges, it's useless to consider the nodes:
func (graph *MyGraph) Distance(node1,node2 Node) int {
return 1
}
As that thread notes, there is a valid argument to only allow parameters named as _ if they're unused (e.g. Distance(_,_ Node)), but at this point it's too late due to the Go 1 future-compatibility guarantee. As also mentioned, a possible objection to that anyway is that parameters, even if unused, can implicitly provide documentation.
In short: there's no concrete, specific answer, other than that they simply made an ultimately arbitrary (but still educated) determination that unused parameters are more important and useful than unused local variables and imports. If there was once a strong design reason, it's not documented anywhere.
The main reason is to be able to implement interfaces that dictate specific methods with specific parameters, even if you don't use all of them in your implementation. This is detailed in #Jsor's answer.
Another good reason is that unused (local) variables are often the result of a bug or the use of a language feature (e.g. use of short variable declaration := in a block, unintentionally shadowing an "outer" variable) while unused function parameters never (or very rarely) are the result of a bug.
Another reason can be to provide forward compatibility. If you release a library, you can't change or extend the parameter list without breaking backward compatibility (and in Go there is no function overloading: if you want 2 variants with different parameters, their names must be different too).
You may provide an exported function or method and add extra - not yet used - or optional parameters (e.g. hints) to it in the spirit that you may use them in a future version / release of your library.
Doing so early will give you the benefit that others using your library won't have to change anything in their code.
Let's see an example:
You want to create a formatting function:
// FormatSize formats the specified size (bytes) to a string.
func FormatSize(size int) string {
return fmt.Sprintf("%d bytes", size)
}
You may as well add an extra parameter right away:
// FormatSize formats the specified size (bytes) to a string.
// flags can be used to alter the output format. Not yet used.
func FormatSize(size int, flags int) string {
return fmt.Sprintf("%d bytes", size)
}
Then later you may improve your library and your FormatSize() function to support the following formatting flags:
const (
FlagAutoUnit = 1 << iota // Automatically format as KB, MB, GB etc.
FlagSI // Use SI conversion (1000 instead of 1024)
FlagGroupDecimals // Format number using decimal grouping
)
// FormatSize formats the specified size (bytes) to a string.
// flags can be used to alter the output format.
func FormatSize(size int, flags int) string {
var s string
// Check flags and format accordingly
// ...
return s
}