W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2020

Re: draft-ietf-httpbis-header-structure-15, Serializing a Decimal

From: Kari Hurtta <hurtta-ietf@elmme-mailer.org>
Date: Thu, 30 Jan 2020 07:52:29 +0200 (EET)
To: Mark Nottingham <mnot@mnot.net>
CC: Kari Hurtta <hurtta-ietf@elmme-mailer.org>, HTTP Working Group <ietf-http-wg@w3.org>, Poul-Henning Kamp <phk@varnish-cache.org>
Message-Id: <20200130055238.1CFF91F35@welho-filter3.welho.com>
> Hi Kari,
> 
> Very good catch; I noticed this a while back but forgot to file an issue. Now:
>   https://github.com/httpwg/http-extensions/issues/1043

| ... still isn't right, as Kari noticed. E.g., a value of 1.9944 won't get rounded to 2.0.

You mean:  E.g., a value of 1.9998  won't get rounded to 2.0.

Algorithm produces  three digits for fractional component.

And 1.994 is correct rounding for 1.9944, I think.

 
> Cheers,
> 

Another thing what can be issue when rounding is applied.

> > On 30 Jan 2020, at 5:17 am, Kari Hurtta <hurtta-ietf@elmme-mailer.org> wrote:
> > 
> > 4.1.5.  Serializing a Decimal
> > https://tools.ietf.org/html/draft-ietf-httpbis-header-structure-15#section-4.1.5
> > 
> > 
> > |   Given a decimal_number as input_decimal, return an ASCII string
> > |   suitable for use in a HTTP header value.
> > |
> > |   1.  Let output be an empty string.
> > |
> > |   2.  If input_decimal is less than (but not equal to) 0, append "-" to
> > |       output.
> > |
> > |   3.  Append input_decimal's integer component represented in base 10
> > |       (using only decimal digits) to output; if it is zero, append "0".
> > |
> > |   4.  If the number of characters appended in the previous step is
> > |       greater than 12, fail serialisation.

A decimal_number  number can be  -999 999 999 999.9997 or 999 999 999 999.9997

In step 3 there is  digits 999999999999 so no failure on step 4


> > |   5.  Append "." to output.
> > |
> > |   6.  If input_decimal's fractional component is zero, append "0" to
> > |       output.
> > |
> > |   7.  Else if input_decimal's fractional component has up to three
> > |       digits, append them represented in base 10 (using only decimal
> > |       digits) to output.
> > |
> > |   8.  Otherwise, append the first three digits of input_decimal's
> > |       fractional component (represented in base 10, using only decimal
> > |       digits) to output, rounding the final digit to the nearest value,
> > |       or to the even value if it is equidistant.
> > |
> > |   9.  Return output.

First theree digits for fractional component are 999

"Correct" rounding for  -999 999 999 999.9997  is -1 000 000 000 000.0 and
for 999 999 999 999.9997 is 1 000 000 000 000.0, but there are 13 digits 
on integer component.

That is "solved" if 
when decimal_number < -999 999 999 999.999 or
decimal_number > 999 999 999 999.999, fractional component
is truncated to 3 digits and rounding is not tried.

/ Kari Hurtta
Received on Thursday, 30 January 2020 05:52:43 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 30 January 2020 05:52:45 UTC