I am currently working on a high-performance Vector/Matrix Ruby gem C extension, as I find the built-in implementation cumbersome and not ideal for most cases that I have personally encountered, as well as lacking in other areas.
My first approach was implementing in Ruby as a subclass of Fiddle::CStructEntity, as a goal is to make them optimized for interop without need for conversion (such as passing to native OpenGL functions). Implementing in C offers a great benefit for the math, but I ran into a roadblock when trying to implement a minor function.
I wished to have a method return a Fiddle::Pointer to the struct (basically a pointer to Rdata->data. I wished to return an actual Fiddle::Pointer object. Returning an integer address, packed string, etc. is trivial, and using that could easily be extended in a Ruby method to convert to a Fiddle::Pointer like this:
def ptr
# Assume address is an integer address returned from C
Fiddle::Pointer.new(self.address, self.size)
end
This kind of opened up a question to me, and that is it possible to to even do such from C? Fiddle is not part of the core, library, it is part of the standard lib, and as such, is really just an extension itself.
The problem is trivial, and can be easily overcome with a couple lines of Ruby code as demonstrated above, but was more curious if returning a Fiddle object was even possible from a C extension without hacks? I was unable to find any examples of this being done, and as always when it comes to the documentation involving Fiddle, it is quite basic and does not explain much.
The solution for this is actually rather simple, though admittedly not as elegant or clean of a solution I was hoping to discover.
There are possibly more elaborate ways to go about this by including the headers for Fiddle, and building against it, but this was not really a viable solution, as I didn't want to restrict my C extension to only work with Ruby 2.0+, and would be perfectly acceptable to simply omit the method in the event Ruby version was less than 2.0.
First I include version.h, which gives access defines the macro RUBY_API_VERSION_MAJOR, which is all I really need to know in regards to whether or not Fiddle will be present or not.
This will be an abbreviated version to simply show how to get the Fiddle::Pointer class as a VALUE, and to create an instance.
#if RUBY_API_VERSION_MAJOR >= 2
rb_require("fiddle");
VALUE fiddle = rb_const_get(rb_cObject, rb_intern("Fiddle"));
rb_cFiddlePointer = rb_const_get(fiddle, rb_intern("Pointer"));
#endif
In this example, the class is stored in rb_cFiddlePointer, which can then be used to create and return a Fiddle::Pointer object from C.
// Get basic data about the struct
struct RData *rdata = RDATA(self);
VALUE *args = xmalloc(sizeof(VALUE) * 2);
// Set the platform pointer-size address (could just use size_t here...)
#if SIZEOF_INTPTR_T == 4
args[0] = LONG2NUM((long) rdata->data);
#elif SIZEOF_INTPTR_T == 8
args[0] = LL2NUM((long long) rdata->data);
#else
args[0] = INT2NUM(0);
#endif
// Get size of structure
args[1] = INT2NUM(SIZE_OF_YOUR_STRUCTURE);
VALUE ptr = rb_class_new_instance(2, args, rb_cFiddlePointer);
xfree(args);
return ptr;
After linking the function to an actual Ruby method, you can then call it to get a sized pointer to the internal structure in memory.
I am new to using Halide and I am playing around with implementing algorithms first. I am trying to write a function which, depending on the value of the 8 pixels around it, either skips to the next pixel or does some processing and then moves on to the next pixel. When trying to write this I get the following compiler error:
84:5: error: value of type 'Halide::Expr' is not contextually convertible to 'bool'
if(input(x,y) > 0)
I have done all the tutorials and have seen that the select function is an option, but is there a way to either compare the values of a function or store them somewhere?
I also may be thinking about this problem wrong or might not be implementing it with the right "Halide mindset", so any suggestions would be great. Thank you in advance for everything!
The underlying issue here is that, although they are syntactically interleaved, and Halide code is constructed by running C++ code, Halide code is not C++ code and vice versa. Halide code is entirely defined by the Halide::* data structures you build up inside Funcs. if is a C control flow construct; you can use it to conditionally build different Halide programs, but you can't use it inside the logic of the Halide program (inside an Expr/Func). select is to Halide (an Expr which conditionally evaluates to one of two values) as if/else is to C (a statement which conditionally executes one of two sub-statements).
Rest assured, you're hardly alone in having this confusion early on. I want to write a tutorial specifically addressing how to think about staged programming inside Halide.
Until then, the short, "how do I do what I want" answer is as you suspected and as Khouri pointed out: use a select.
Since you've provided no code other than the one line, I'm assuming input is a Func and both x and y are Vars. If so, the result of input(x,y) is an Expr that you cannot evaluate with an if, as the error message indicates.
For the scenario that you describe, you might have something like this:
Var x, y;
Func input; input(x,y) = ...;
Func output; output(x,y) = select
// examine surrounding values
( input(x-1,y-1) > 0
&& input(x+0,y-1) > 0
&& ...
&& input(x+1,y+1) > 0
// true case
, ( input(x-1,y-1)
+ input(x+0,y-1)
+ ...
+ input(x+1,y+1)
) / 8
// false case
, input(x,y)
);
Working in Halide definitely requires a different mindset. You have to think in a more mathematical form. That is, a statement of a(x,y) = b(x,y) will be enforced for all cases of x and y.
Algorithm and scheduling should be separate, although the algorithm may need to be tweaked to allow for better scheduling.
I'm trying to implement a fork-merge parser for C using Java. I need to fork the parser whenever I find an #if directive. For example:
int x = #if 3; #else 4; #endif
The above statement should be parsed as follows:
First I create a new parser for #if and read-in everything under #if statement, in the above case, reading value 3 directly wold throw a syntax error, in that case I should read back all the tokens that were already read. How do I do this?
Well, you could take any standard parser, and replicate its state on encountering an #if. Then one version goes down the fork of the #if, the other down the fork of the #else. (Doesn't have a #else? Pretend you saw #if cond lexemes #else #endif).
Life gets messy if the preprocessor conditionals don't occur in nice places:
void p()
{ if (x<2)
{ y =
#if cond
3; } else { x=
#else
7;
#endif
}
The parse can end up in really different states at the end of each conditional subpart. A consequence: you need to expand preprocessor conditionals so they include clean language structures.
A serious problem with this approach, is that if you fork every time you see #if, you end up with 2^N forks for N #ifs. It is easy to find C code with dozens of conditionals; 2^24 --> 16 million so this gets out of hand fast.
So you need a way to merge the parses back together again when you hit the #endif. That's not so easy; we have done this with a GLR parser but the answer is very complicated and won't fit easily here. This technical paper discusses how to do this merge for LR parsers: http://www.lirmm.fr/~ducour/Doc-objets/ECOOP2012/PLDI/pldi/p323.pdf
There's a second complication: imagine
#if cond1
stuff1
#else
#if cond2
stuff4
#else
stuff5
#endif
#endif
Now you need to fork parsers inside parsers. Worse, stuff4 has a condition, which is the conjunction of cond1 and cond2, but stuff5 has a condition of cond1 & ~ cond2. Realistically, you're going to need to compute and retain the condition under which each parse (and generated subtree) occurs. You'll need some kind of symbolic condition computation to do this, and you'll need to handle the case where the composed condition is entirely false specially (just skip the content). Interestingly, our GLR solution and the above technical paper both agree that using BDDs for this is a good idea.
If you want to do refactoring, you'll need to determine the meaning of names in the presence of conditionals:
#if cond1
float x;
#else
char* x;
#endif
...
x=
#if cond1
3.7
#else
"foobar"
#endif
;
This requires having a symbol table that carries conditional information with the symbols. See my technical paper http://www.rcost.unisannio.it/mdipenta/papers/scam2002.pdf for details on how to approach this.
To do refactoring, you're going to need control and data flow analysis on top of all this.
Check out my bio for a tool where we are attempting to do all this. We have the conditional parsing part done right, we think. The rest is still up in the air.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
When have you run into syntax that might be dated, never used or just plain obfuscated that you couldn't understand for the life of you.
For example, I never knew that comma is an actual operator in C. So when I saw the code
if(Foo(), Bar())
I just about blew a gasket trying to figure out what was going on there.
I'm curious what little never-dusted corners might exist in other languages.
C++'s syntax for a default constructor on a local variable. At first I wrote the following.
Student student(); // error
Student student("foo"); // compiles
This lead me to about an hour of reading through a cryptic C++ error message. Eventually a non-C++ newbie dropped by, laughed and pointed out my mistake.
Student student;
This is always jarring:
std::vector <std::vector <int> >
^
mandatory space.
When using the System.DirectoryServices name space to bind to an ADAM (Active Directory Application Mode; now called AD LDS, I think), I lost an entire day trying to debug this simple code:
DirectoryEntry rootDSE = new DirectoryEntry(
"ldap://192.168.10.78:50000/RootDSE",
login,
password,
AuthenticationTypes.None);
When I ran the code, I kept getting a COMException with error 0x80005000, which helpfully mapped to "Unknown error."
I could use the login and password and bind to the port via ADSI Edit. But this simple line of code didn't work. Bizarre firewall permission? Something screwed in configuration? Some COM object not registered correctly? Why on earth wasn't it working?
The answer? It's LDAP://, not ldap://.
And this is why we drink.
C++
class Foo
{
// Lots of stuff here.
} bar;
The declaration of bar is VERY difficult to see. More commonly found in C, but especially annoying in C++.
Perl's syntax caused me a bad day a while ago:
%table = {
foo => 1,
bar => 2
};
Without proper warnings (which are unavailable on the platform I was using), this creates a one-element hash with a key as the given hash reference and value undef. Note the subtle use of {}, which creates a new hash reference, and not (), which is an array used to populate the %table hash.
I was shocked Python's quasi-ternary operator wasn't a syntax error the first time I saw it:
X if Y else Z
This is stupid and common, but this syntax:
if ( x = y ) {
// do something
}
Has caught me about three times in the past year in a couple of different languages. I really like the R language's convention of using <- for assignment, like this:
x <- y
If the x = y syntax were made to mean x == y, and x <- y to mean assignment, my brain would make a smoother transition to and from math and programming.
C/C++'s bitvector syntax. The worst part about this is trying to google for it simply based on the syntax.
struct C {
unsigned int v1 : 12;
unsigned int v2 : 1;
};
C#'s ?? operator threw me for a loop the first time I saw it. Essentially it will return the LHS if it's non-null and the RHS if the LHS is null.
object bar = null;
object foo = bar ?? new Student(); // gets new Student()
Powershell's function calling semantics
function foo() {
params ($count, $name);
...
}
foo (5, "name")
For the non powershellers out there. This will work but not how you expect it to. It actually creates an array and passes it as the first argument. The second argument has no explicit value. The correct version is
foo 5 "name"
The first time I saw a function pointer in C++ I was confused. Worse, because the syntax has no key words, it was really hard to look up. What exactly does one type into a search engine for this?
int (*Foo)(float, char, char);
I ended up having to ask the local C++ guru what it was.
VB's (yeah yeah, I have to use it) "And" keyword - as in:
If Object IsNot Nothing And Object.Property Then
See that Object.Property reference, after I've made sure the object isn't NULL? Well, VB's "And" keyword * does * not * block * further * evaluation and so the code will fail.
VB does have, however, another keyword - AndAlso:
If Object IsNot Nothing AndAlso Object.Property Then
That will work as you'd expect and not explode when run.
I was once very confused by some C++ code that declared a reference to a local variable, but never used it. Something like
MyLock &foo;
(Cut me some slack on the syntax, I haven't done C++ in nearly 8 years)
Taking that seemingly unused variable out made the program start dying in obscure ways seemingly unrelated to this "unused" variable. So I did some digging, and found out that the default ctor for that class grabbed a thread lock, and the dtor released it. This variable was guarding the code against simultaneous updates without seemingly doing anything.
Javascript: This syntax ...
for(i in someArray)
... is for looping through arrays, or so I thought. Everything worked fine until another team member dropped in MooTools, and then all my loops were broken because the for(i in ...) syntax also goes over extra methods that have been added to the array object.
Had to translate some scientific code from old FORTRAN to C. A few things that ruined my day(s):
Punch-card indentation. The first 6 characters of every line were reserved for control characters, goto labels, comments, etc:
^^^^^^[code starts here]
c [commented line]
Goto-style numbering for loops (coupled with 6 space indentation):
do 20, i=0,10
do 10, j=0,10
do_stuff(i,j)
10 continue
20 continue
Now imagine there are multiple nested loops (i.e., do 20 to do 30) which have no differentiating indentation to know what context you are in. Oh, and the terminating statements are hundreds of lines away.
Format statement, again using goto labels. The code wrote to files (helpfully referred to by numbers 1,2,etc). To write the values of a,b,c to file we had:
write (1,51) a,b,c
So this writes a,b,c to file 1 using a format statement at the line marked with label 51:
51 format (f10.3,f10.3,f10.3)
These format lines were hundreds of lines away from where they were called. This was complicated by the author's decision to print newlines using:
write (1,51) [nothing here]
I am reliably informed by a lecturer in the group that I got off easy.
C's comma operator doesn't seem very obscure to me: I see it all the time, and if I hadn't, I could just look up "comma" in the index of K&R.
Now, trigraphs are another matter...
void main() { printf("wat??!\n"); } // doesn't print "wat??!"
Wikipedia has some great examples, from the genuinely confusing:
// Will the next line be executed????????????????/
a++;
to the bizarrely valid:
/??/
* A comment *??/
/
And don't even get me started on digraphs. I would be surprised if there's somebody here who can fully explain C's digraphs from memory. Quick, what digraphs does C have, and how do they differ from trigraphs in parsing?
Syntax like this in C++ with /clr enabled. Trying to create a Managed Dictionary object in C++.
gcroot<Dictionary<System::String^, MyObj^>^> m_myObjs;
An oldie:
In PL/1 there are no reserved words, so you can define variables, methods, etc. with the same name as the language keywords.
This can be a valid line of code:
IF ELSE THEN IF ELSE THEN
(Where ELSE is a boolean, and IF and THEN are functions, obviously.)
Iif(condition, expression, expression) is a function call, not an operator.
Both sides of the conditional are ALWAYS evaluated.
It always ruines my day if I have to read/write some kind of Polish notation as used in a lot of HP calculators...
PHP's ternary operator associates left to right. This caused me much anguish one day when I was learning PHP. For the previous 10 years I had been programming in C/C++ in which the ternary operator associates right to left.
I am still a little curious as to why the designers of PHP chose to do that when, in many other respects, the syntax of PHP matches that C/C++ fairly closely.
EDIT: nowadays I only work with PHP under duress.
Not really obscure, but whenever I code too much in one language, and go back to another, I start messing up the syntax of the latter. I always chuckle at myself when I realize that "#if" in C is not a comment (but rather something far more deadly), and that lines in Python do not need to end in a semicolon.
While performing maintentnace on a bit of C++ code I once spotted that someone had done something like this:
for (i=0; i<10; i++)
{
MyNumber += 1;
}
Yes, they had a loop to add 1 to a number 10 times.
Why did it ruin my day? The perpetrator had long since left, and I was having to bug fix their module. I thought that if they were doing something like this, goodness knows what else I was going to encounter!
AT&T assembler syntax >:(
This counter-intuitive, obscure syntax has ruined many of my days, for example, the simple Intel syntax assembly instruction:
mov dword es:[ebp-5], 1 /* Cool, put the value 1 into the
* location of ebp minus five.
* this is so obvious and readable, and hard to mistake
* for anything else */
translates into this in AT&T syntax
movl $1, %es:-4(%ebp) /* huh? what's "l"? 4 bytes? 8 bytes? arch specific??
* wait, why are we moving 1 into -4 times ebp?
* or is this moving -4 * ebp into memory at address 0x01?
* oh wait, YES, I magically know that this is
* really setting 4 bytes at ebp-5 to 1!
More...
mov dword [foo + eax*4], 123 /* Intel */
mov $123, foo(, %eax, 4) /* AT&T, looks like a function call...
* there's no way in hell I'd know what this does
* without reading a full manual on this syntax */
And one of my favorites.
It's as if they took the opcode encoding scheme and tried to incorporate it into the programming syntax (read: scale/index/base), but also tried to add a layer of abstraction on the data types, and merge that abstraction into the opcode names to cause even more confusion. I don't see how anyone can program seriously with this.
In a scripting language (Concordance Programming Language) for stand alone database software (Concordance) used for litigation document review, arrays were 0 indexed while (some) string functions were 1 indexed. I haven't touched it since.
This. I had my run in with it more then once.
GNU extensions are often fun:
my_label:
unsigned char *ptr = (unsigned char *)&&my_label;
*ptr = 5; // Will it segfault? Finding out is half the fun...
The syntax for member pointers also causes me grief, more because I don't use it often enough than because there's anything really tricky about it:
template<typename T, int T::* P>
function(T& t)
{
t.*P = 5;
}
But, really, who needs to discuss the obscure syntax in C++? With operator overloading, you can invent your own!