www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Why ByChunk is not regnize by st.range.takeOne and why Section has no

reply "bioinfornatics" <bioinfornatics fedoraproject.org> writes:
Dear,
Some problem to build this code: 
http://fpaste.org/147327/75948141/


$ ldc2 fasta_test.d
/usr/include/d/std/range.d(3605): Error: template std.array.save
cannot deduce function from argument types !()(ByChunk),
candidates are:
/usr/include/d/std/array.d(554):        std.array.save(T)(T[] a)
fasta_test.d(46): Error: template instance
std.range.takeOne!(ByChunk) error instantiating
fasta_test.d(88):        instantiated from here: Parser!(Fasta,
ByChunk)
fasta_test.d(79): Error: type Section!((words) => words[0] ==
'>', (words) => words[0] == '\x0a') has no value
fasta_test.d(50): Error: template std.algorithm.countUntil cannot
deduce function from argument types !(__lambda3)(ubyte[]),
candidates are:
/usr/include/d/std/algorithm.d(5406):
std.algorithm.countUntil(alias pred = "a == b", R, Rs...)(R
haystack, Rs needles) if (isForwardRange!R && Rs.length > 0 &&
isForwardRange!(Rs[0]) == isInputRange!(Rs[0]) &&
is(typeof(startsWith!pred(haystack, needles[0]))) && (Rs.length
== 1 || is(typeof(countUntil!pred(haystack,
needles[1..__dollar])))))

Thanks for your help
Nov 02 2014
parent reply =?UTF-8?B?QWxpIMOHZWhyZWxp?= <acehreli yahoo.com> writes:
On 11/02/2014 04:58 PM, bioinfornatics wrote:


 Dear,
 Some problem to build this code: http://fpaste.org/147327/75948141/


 $ ldc2 fasta_test.d
 /usr/include/d/std/range.d(3605): Error: template std.array.save
 cannot deduce function from argument types !()(ByChunk),
 candidates are:


I think that is due to the too permissive template constraint of takeOne 
in phobos/std/range.d. Even though it say "isInputRange!R", it clearly 
needs "isForwardRange!R" because it calls .save() on that range:

      property auto save() { return Result(_source.save, empty); }

So, the following is wrong:

   auto takeOne(R)(R source) if (isInputRange!R)

It should be:

   auto takeOne(R)(R source) if (isForwardRange!R)

If you modify range.d locally to require ForwardRange, you will see that 
your following call will get an error message because byChunk is not a 
ForwardRange:

         // Take a piece of byte from current file
         ubyte[] buffer =  takeOne(inputRange);

I think this issue is already reported as a part of the following bug:

   https://issues.dlang.org/show_bug.cgi?id=9724

Ali
Nov 02 2014
parent reply "bioinfornatics" <bioinfornatics fedoraproject.org> writes:
On Monday, 3 November 2014 at 06:43:50 UTC, Ali Çehreli wrote:

 On 11/02/2014 04:58 PM, bioinfornatics wrote:


 Dear,
 Some problem to build this code:


 http://fpaste.org/147327/75948141/

 $ ldc2 fasta_test.d
 /usr/include/d/std/range.d(3605): Error: template


 std.array.save

 cannot deduce function from argument types !()(ByChunk),
 candidates are:


 I think that is due to the too permissive template constraint 
 of takeOne in phobos/std/range.d. Even though it say 
 "isInputRange!R", it clearly needs "isForwardRange!R" because 
 it calls .save() on that range:

      property auto save() { return Result(_source.save, empty); 
 }

 So, the following is wrong:

   auto takeOne(R)(R source) if (isInputRange!R)

 It should be:

   auto takeOne(R)(R source) if (isForwardRange!R)

 If you modify range.d locally to require ForwardRange, you will 
 see that your following call will get an error message because 
 byChunk is not a ForwardRange:

         // Take a piece of byte from current file
         ubyte[] buffer =  takeOne(inputRange);

 I think this issue is already reported as a part of the 
 following bug:

   https://issues.dlang.org/show_bug.cgi?id=9724

 Ali


Ok but I do not see why to use save method for a takeOne function
is possible to write this function without to use it
Nov 03 2014
next sibling parent Jonathan M Davis via Digitalmars-d-learn writes:
On Monday, November 03, 2014 21:03:51 bioinfornatics via Digitalmars-d-learn
wrote:

 Ok but I do not see why to use save method for a takeOne function
 is possible to write this function without to use it


It looks like the idea was that beacause you know that there's only one
element, you can make it random-access range with opSlice and popBack and the
whole shebang. And that works great if the range is at least a forward range,
but it currently falls apart with an input range, because while you can easily
emulate stuff like back and opIndex by forwarding it to front, save can't be
emulated so easily. So, either front needs to be accessed and saved in the
Result struct rather than than simply having it contain the range that it's
forwarding to, or it needs to have another branch of the static if for
handling input ranges separately. Though to be honest, I'm not sure that
using takeOne makes a lot of sense over take in that case. You don't really
get a performance boost, and since you can't add on the random-access range
capabilities without making it eager and saving the result of front (which is
an iffy thing to do with some input ranges), it's really not adding anything
over plain take. I'd honestly favor just fixing takeOne's template constraint
so that it checked for forward ranges. The only reason that I see to make
takeOne work with input ranges is simply so that no one will be surprised when
it doesn't work with an input range. But that's not really adding any useful
functionality as far as I can tell. Though maybe there's something that I'm
missing.

- Jonathan M Davis
Nov 03 2014
prev sibling parent "bioinfornatics" <bioinfornatics fedoraproject.org> writes:
On Monday, 3 November 2014 at 21:03:51 UTC, bioinfornatics wrote:

 On Monday, 3 November 2014 at 06:43:50 UTC, Ali Çehreli wrote:

 On 11/02/2014 04:58 PM, bioinfornatics wrote:


 Dear,
 Some problem to build this code:


 http://fpaste.org/147327/75948141/

 $ ldc2 fasta_test.d
 /usr/include/d/std/range.d(3605): Error: template


 std.array.save

 cannot deduce function from argument types !()(ByChunk),
 candidates are:


 I think that is due to the too permissive template constraint 
 of takeOne in phobos/std/range.d. Even though it say 
 "isInputRange!R", it clearly needs "isForwardRange!R" because 
 it calls .save() on that range:

     property auto save() { return Result(_source.save, empty); 
 }

 So, the following is wrong:

  auto takeOne(R)(R source) if (isInputRange!R)

 It should be:

  auto takeOne(R)(R source) if (isForwardRange!R)

 If you modify range.d locally to require ForwardRange, you 
 will see that your following call will get an error message 
 because byChunk is not a ForwardRange:

        // Take a piece of byte from current file
        ubyte[] buffer =  takeOne(inputRange);

 I think this issue is already reported as a part of the 
 following bug:

  https://issues.dlang.org/show_bug.cgi?id=9724

 Ali


 Ok but I do not see why to use save method for a takeOne 
 function
 is possible to write this function without to use it


Example:

auto takeOne(R)(ref R source) if (isInputRange!R)
{
      static if (hasSlicing!R)
      {
          return source[0 .. !source.empty];
      }
      else
      {
          typeof(R.front) r;
          if(!source.empty){
              r = source.front;
              source.popFront();
          }

          return r;
      }
}
Nov 03 2014