Differences

This shows you the differences between two versions of the page.

--- ada.calendar.clock_to_string [2012/01/02 20:24]
thomaslocke [One Solution]
+++ ada.calendar.clock_to_string [2012/01/03 21:48] (current)
thomaslocke [A Full Program]
@@ Line 1: / Line 1: @@
 =====  Introduction  =====
-While messing around with [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3741|Ada.Calendar]], a need to output the return value of [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Clock]] as a [[http://adaic.org/standards/05rm/html/RM-3-6-3.html#I1889|String]], took me on an adventure in the land of [[http://adaic.org/standards/05rm/html/RM-11-4-2.html|pragma Assert]], [[http://adaic.org/standards/05rm/html/RM-13-3.html#I4496|the Address attribute]], [[http://adaic.org/standards/05rm/html/RM-13-3.html#I4513|the 'Size attribute]] and the fact that not all Ada compilers are created equal.
+While messing around with [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3741|Ada.Calendar]], a need to output the return value of [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Clock]] as a [[http://adaic.org/standards/05rm/html/RM-3-6-3.html#I1889|String]], took me on an adventure in the land of [[http://adaic.org/standards/05rm/html/RM-11-4-2.html|pragma Assert]], [[http://adaic.org/standards/05rm/html/RM-13-3.html#I4496|the 'Address attribute]], [[http://adaic.org/standards/05rm/html/RM-13-3.html#I4513|the 'Size attribute]], [[http://www.adaic.org/resources/add_content/standards/05rm/html/RM-13-9.html|Ada.Unchecked_Conversion]] and the fact that not all Ada compilers are created equal.
 =====  Caveat  =====
-The following is an example of how to do something. It's not meant to be used in any real capacity. If you need a portable [[http://adaic.org/standards/05rm/html/RM-3-6-3.html#I1889|String]] representation of [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Clock]], then you're better off using the many nice features of [[http://adaic.org/standards/05rm/html/RM-9-6-1.html#I3773|Ada.Calendar.Formatting]].
+The following are examples of how to do something. It's not meant to be used in any real capacity. If you need a portable [[http://adaic.org/standards/05rm/html/RM-3-6-3.html#I1889|String]] representation of [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Clock]], then you're better off using the many nice features of [[http://adaic.org/standards/05rm/html/RM-9-6-1.html#I3773|Ada.Calendar.Formatting]].
 =====  The Problem  =====
@@ Line 11: / Line 11: @@
 The problem of outputting the return value from [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Clock]] as a string is one of privacy. The returned type, [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Time]], is declared //private//. Also it's implementation defined:
-<blockquote>
-The time base associated with the type Time of package Calendar is implementation defined. The function Clock of package Calendar returns a value representing the current time for this time base. The implementation-defined value of the named number System.Tick (see 13.7) is an approximation of the length of the real-time interval during which the value of Calendar.Clock remains constant.
+> The time base associated with the type Time of package Calendar is implementation defined. The function Clock of package Calendar returns a value representing the current time for this time base. The implementation-defined value of the named number System.Tick (see 13.7) is an approximation of the length of the real-time interval during which the value of Calendar.Clock remains constant.
-</blockquote>
 So we can't see it, and we don't know how it's implemented. But there are ways around these issues.
-=====  One Solution  =====
+===== One Solution =====
-For my experiment I've used the [[http://libre.adacore.com/libre/|GNAT GPL 2010]] compiler, and in this compiler the [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Time]] type is declared as:
+In this first solution, we're going to make use of the //'Address// attribute. For my experiment I've used the [[http://libre.adacore.com/libre/|GNAT GPL 2010]] compiler, and in this compiler the [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Time]] type is declared as:
 <code ada>
@@ Line 29: / Line 28: @@
 <code ada>
-function Time_Image
+function Time_Image_One
   (Item : in Ada.Calendar.Time)
    return String
 is
-   type Time is mod 2 ** 64;
+   type New_Time is mod 2 ** 64;
-   pragma Assert (Time'Size = Item'Size);
+   pragma Assert (New_Time'Size = Item'Size);
-   Time_Stamp : Time;
+   Time_Stamp : New_Time;
    for Time_Stamp'Address use Item'Address;
 begin
-   return Trim (Source => Time'Image (Time_Stamp),
+   return Trim (Source => New_Time'Image (Time_Stamp),
                 Side   => Ada.Strings.Left);
-end Time_Image;
+end Time_Image_One;
 </code>
@@ Line 49: / Line 48: @@
 <code ada>
-type Time is mod 2 ** 64;
+type New_Time is mod 2 ** 64;
 </code>
-We declare a new //Time// type with the same size as the [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Time]] type (as defined by the GNAT GPL 2010 compiler). I've used //mod 2 <nowiki>**</nowiki> 64// instead of //range -2 ** 63 .. +2 ** 63 -1//, but that's mostly because I don't care about the sign. You can just as well go with //range -2 ** 63 .. +2 ** 63 -1//. The important thing to bear in mind, is that the two types should have the same size. Which brings us to this:
+We declare a new //Time// type with the same size as the [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Time]] type (as defined by the GNAT GPL 2010 compiler). I've used //mod 2 <nowiki>**</nowiki> 64// instead of //range -2 <nowiki>**</nowiki> 63 .. +2 <nowiki>**</nowiki> 63 -1//, but that's mostly because I don't care about the sign. You can just as well go with //range -2 <nowiki>**</nowiki> 63 .. +2 <nowiki>**</nowiki> 63 -1//. The important thing to bear in mind, is that the two types should have the same size. Which brings us to this:
 <code ada>
-pragma Assert (Time'Size = Item'Size);
+pragma Assert (New_Time'Size = Item'Size);
 </code>
@@ Line 65: / Line 64: @@
 <code ada>
-Time_Stamp : Time;
+Time_Stamp : New_Time;
 for Time_Stamp'Address use Item'Address;
 </code>
@@ Line 74: / Line 73: @@
 <code ada>
-return Trim (Source => Time'Image (Time_Stamp),
+return Trim (Source => New_Time'Image (Time_Stamp),
              Side   => Ada.Strings.Left);
 </code>
@@ Line 80: / Line 79: @@
 And voila! We have a neat [[http://adaic.org/standards/05rm/html/RM-3-6-3.html#I1889|String]] representation of the [[http://adaic.org/standards/05rm/html/RM-9-6.html#I3747|Ada.Calendar.Clock]] return value. Whether or not this is useful for anything, well, I'll leave that decision to the reader.
-=====  Thanks  =====
+===== Another Solution =====
+The above solution is not the only way to accomplish the goal of outputting the return value of //Ada.Calendar.Clock//. There's another, and probably better, method: Using [[http://www.adaic.org/resources/add_content/standards/05rm/html/RM-13-9.html|Ada.Unchecked_Conversion]]. Here's how the above //Time_Image_One// function would look if we used //Ada.Unchecked_Conversion// instead:
+<code ada>
+function Time_Image_Two
+  (Item : in Ada.Calendar.Time)
+   return String
+is
+   type New_Time is mod 2 ** 64;
+   pragma Assert (New_Time'Size = Ada.Calendar.Time'Size);
+   function To_Time
+   is new Ada.Unchecked_Conversion (Source => Ada.Calendar.Time,
+                                    Target => New_Time);
+   Time_Stamp : constant New_Time := To_Time (Item);
+begin
+   return Trim (Source => New_Time'Image (Time_Stamp),
+                Side   => Ada.Strings.Left);
+end Time_Image_Two;
+</code>
+A great boon of this method over the first one, is this fantastic warning message the GNAT compiler will dump on you when you compile the program:
+  warning: representation of Time values may change between GNAT versions
+You're not going to get that with the first method, and in my humble opinion that warning alone makes it worthwhile to use //Ada.Unchecked_Conversion// instead of the //'Address// attribute used in the first solution.
+===== The Full Program =====
+Lets end with a little program that utilizes both solutions to output the value of //Ada.Calendar.Clock//:
+<code ada>
+with Ada.Calendar;
+with Ada.Strings.Fixed;
+with Ada.Text_IO;
+with Ada.Unchecked_Conversion;
+procedure Time_Image_Test is
+   use Ada.Strings.Fixed;
+   use Ada.Text_IO;
+   ----------------------
+   --  Time_Image_One  --
+   ----------------------
+   function Time_Image_One
+     (Item : in Ada.Calendar.Time)
+      return String
+   is
+      type New_Time is mod 2 ** 64;
+      pragma Assert (New_Time'Size = Item'Size);
+      Time_Stamp : New_Time;
+      for Time_Stamp'Address use Item'Address;
+   begin
+      return Trim (Source => New_Time'Image (Time_Stamp),
+                   Side   => Ada.Strings.Left);
+   end Time_Image_One;
+   ----------------------
+   --  Time_Image_Two  --
+   ----------------------
+   function Time_Image_Two
+     (Item : in Ada.Calendar.Time)
+      return String
+   is
+      type New_Time is mod 2 ** 64;
+      pragma Assert (New_Time'Size = Ada.Calendar.Time'Size);
+      function To_Time
+      is new Ada.Unchecked_Conversion (Source => Ada.Calendar.Time,
+                                       Target => New_Time);
+      Time_Stamp : constant New_Time := To_Time (Item);
+   begin
+      return Trim (Source => New_Time'Image (Time_Stamp),
+                   Side   => Ada.Strings.Left);
+   end Time_Image_Two;
+   Now : Ada.Calendar.Time := Ada.Calendar.Clock;
+begin
+   Put_Line (Time_Image_One (Now));
+   Put_Line (Time_Image_Two (Now));
+end Time_Image_Test;
+</code>
+Elegant stuff!
+===== Thanks =====
-I'd like to extend my thanks to [[http://edb.jacob-sparre.dk/Ada/|Jacob Sparre]] for gently pushing me in the right direction. As usual the [[http://freenode.net/|freenode #ada IRC channel]] proved an invaluable help in learning something new.
+I'd like to extend my thanks to [[http://edb.jacob-sparre.dk/Ada/|Jacob Sparre]] for gently pushing me in the right direction in regards to solution one, and [[http://identi.ca/simonj|Simon J. Wright]] for giving me the second solution. As usual the [[http://freenode.net/|freenode #ada IRC channel]] and the [[http://identi.ca/thomaslocke/all/ada| identi.ca Ada group]] proved an invaluable help in learning something new.

Trace:

Differences

Navigation

Search

Toolbox