Le 13/09/2019 à 16:52, Samuel Gougeon a écrit :
Le 13/09/2019 à 14:22, Stéphane Mottelet a écrit :
However, as I already said it elsewhere, some glitches such as the
following one do occur (see the display of whole x)
--> x=1:0.1:2
x =
1. 1.1 1.2 1.3 1.4 1.5 1.6 1.7000000 1.8 1.9 2.
I agree with Christophe. This output is OK for me. Aestheticism must
be encouraged provided that it does not truncate or downgrade the
information.
About padding every number: Not OK. This would kill one of the assets
of the "v" format: its compacity.
About the fact that 1.7 can't be exactly encoded: It is very
surprising for a so limited decimal number. But OK. I am also quite
surprised that, in this series, only 1.7 can't be exactly encoded.
bitstring allows to see that only 1, 1.5 and 2 are exactly encoded
--> bitstring(1:0.1:2)'
ans =
!0011111111110000000000000000000000000000000000000000000000000000 !
! !
!0011111111110001100110011001100110011001100110011001100110011010 !
! !
!0011111111110011001100110011001100110011001100110011001100110011 !
! !
!0011111111110100110011001100110011001100110011001100110011001101 !
! !
!0011111111110110011001100110011001100110011001100110011001100110 !
! !
!0011111111111000000000000000000000000000000000000000000000000000 !
! !
!0011111111111001100110011001100110011001100110011001100110011010 !
! !
!0011111111111011001100110011001100110011001100110011001100110100 !
! !
!0011111111111100110011001100110011001100110011001100110011001101 !
! !
!0011111111111110011001100110011001100110011001100110011001100110 !
! !
!0100000000000000000000000000000000000000000000000000000000000000 !
So, the discussion holds on the criterion according to which trailing
zeros must be displayed or not.
1. I am wondering about the following, clearly without definitive
opinion. Just a thought:
After format(10), 1.7000000 is displayed if the NEXT figure is not
0, and 1.7 is displayed otherwise.
In other words, this would no longer refer to %eps but to the
format's length.
The issue with this proposal is that we don't have the current
format in mind. If all numbers are displayed in a compact form, we
don't see the display accuracy..
The choice to refer either to %eps or to format() could be
proposed through the preferences.
2. Instead, the discussion could also be about the IEEE rounding
mode. In some occasion, the IEEE rounding mode below %eps has
visible effects on results (there is something about this in
Bugzilla on mailing lists...). Now, i guess that testing with a
hardcoded equivalent of nearfloat() would be too time-consuming.
_______________________________________________
users mailing list
users@lists.scilab.org
https://antispam.utc.fr/proxy/1/c3RlcGhhbmUubW90dGVsZXRAdXRjLmZy/lists.scilab.org/mailman/listinfo/users
--
Stéphane Mottelet
Ingénieur de recherche
EA 4297 Transformations Intégrées de la Matière Renouvelable
Département Génie des Procédés Industriels
Sorbonne Universités - Université de Technologie de Compiègne
CS 60319, 60203 Compiègne cedex
Tel : +33(0)344234688
http://www.utc.fr/~mottelet
_______________________________________________
users mailing list
users@lists.scilab.org
http://lists.scilab.org/mailman/listinfo/users
