inconsistent behavior of float_of_string (compare to "ints"_of_string)

[vicuna]

Original bug ID: 6296
Reporter: @hhugo
Status: closed (set by @xavierleroy on 2017-09-24T15:32:15Z)
Resolution: fixed
Priority: normal
Severity: tweak
Target version: 4.03.0+dev / +beta1
Fixed in version: 4.03.0+dev / +beta1
Category: documentation
Related to: #4323 #5829 #6649 #6701
Monitored by: braibant

Bug description

float_of_string does not understand the octal string representation where other conversion functions do.
float_of_string "0o123" -> fail
(int_/Int64./Int32.)of_string "0o132" -> work

need to document the behavior or to fix it.

also I just discovered (reading strtod manpage) the syntax "0x1p8".

would need either documentation of accepted formats (like is it done with int conversion) or link to documentation of underlying c function

related to #5829

[vicuna]

Comment author: @damiendoligez

I find it strange to expect float_of_string to understand octal integer format.

Still, documenting the accepted formats would be good.

[vicuna]

Comment author: @hhugo

related to ocaml-community/yojson#13

[vicuna]

Comment author: @xavierleroy

Added some documentation for float_of_string, commit [trunk 4d478a0].

Hexadecimal floating-point notation is now officially supported starting with 4.03.

We will not support octal notation. (Nobody does, and octal is a survivance of the past.)