[Octave-bug-tracker] [bug #64949] jsonencode() not supporting single floats

[Jesper PEdersen]

URL:
  <https://savannah.gnu.org/bugs/?64949>

                 Summary: jsonencode() not supporting single floats
                   Group: GNU Octave
               Submitter: jesper_p
               Submitted: Thu 30 Nov 2023 09:46:37 AM UTC
                Category: Octave Function
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Unexpected Error or Warning
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
                 Release: 8.4.0
         Discussion Lock: Any
        Operating System: Any
           Fixed Release: None
         Planned Release: None


    _______________________________________________________

Follow-up Comments:


-------------------------------------------------------
Date: Thu 30 Nov 2023 09:46:37 AM UTC By: Jesper PEdersen <jesper_p>
Passing single-floats to jsonencode() throws an error (it works in Matlab).
E.g.:

octave:1> jsonencode(single(123.456))
error: jsonencode: unsupported type


I've mostly been a user of Octave rather than a contributor so bear with me.
However, I looked into the code and found that I could fix the issue by a
simple modification to jsonencode.cc

diff -r f511a54000f7 libinterp/corefcn/jsonencode.cc
--- a/libinterp/corefcn/jsonencode.cc   Tue Nov 28 22:02:57 2023 -0800
+++ b/libinterp/corefcn/jsonencode.cc   Thu Nov 30 10:41:25 2023 +0100
@@ -75,7 +75,7 @@
   // Possibly write NULL for non-finite values (-Inf, Inf, NaN, NA)
   else if (ConvertInfAndNaN && ! octave::math::isfinite (value))
     writer.Null ();
-  else if (obj.is_double_type ())
+  else if (obj.is_double_type () || obj.is_single_type ())
     writer.Double (value);
   else
     error ("jsonencode: unsupported type");


I am aware that the underlying RapidJSON only supports double which means that
I get some conversion artifacts like below. However, I feel that most users
would prefer this simple solution over the current limitation. Or maybe
someone has an idea of how to avoid below artifacts?

Behaviour after testing the above fix:


octave:1> jsonencode(single(123.456))
ans = 123.45600128173828










    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?64949>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/