Some spec archaeology on number

Hello public-svg-wg,

I'm attempting to do

Fix the definition of number to allow scientific notation in attribute values

from the 22 April SVG telcon


a) we noticed that the current text about <number> wrongly dissallows scientific notation in attribute values. This is a change from both 1.1 and 1.2T which are clear on the subject. See for example SVGT1.2

The current wording in 1.1SE is new, and incorrect.

b) we noticed that the EBNF should allow values like .3 -.4 .5e7 and so on (with no leading numerals before the decimal point). That in turn relates to

Fix the broken grammar for scientific notation in SVG 1.1 and 1.2T

Here is the CVS log of types.html in 1.1SE

Revision 1.5 by jwatt on 19 June 2009 has the cvs log comment

Clarify \<number\> text as decided at today's F2F meeting.

and seems to me to be correct and clarifies 1.1 regarding properties - you can use scientific notation in presentation attributes but not in style sheets (style attributes, style elements, external style sheets)

The cvs log doesn't give an action number, but the minutes from the Raleigh meeting show that no action was given because the edit was done right there at the meeting:

So that change was correct. 

Revision 1.6 by Cameron one day later (11 June 2009) made a bunch of (mostly good) changes but undid jwatt's edits

but erroneously introduced "or in a property's corresponding <a>presentation attribute</a>,"

the cvs log says

Separate decimal and scientific notation numbers.

Make top level productions number and decimal-number and make sure props use decimal-number

which is not linked to any minutes. As the edits took place on day 4 of the Raleigh meeting I looked at the minutes for that day. There is just an IRC log as we were clearing off actions that day

but I don't see 2602 mentioned in there.
The first action given at the f2f on 9 June is 2603, and the last action given on 8 June is 2601.

Revision 1.7, also by Cameron on 13 July 2009 changes it again. The cvs log says

Reversed some of ACTION-2602, which was to separate <number> into two
definitions (one for numbers that allow scientific notation and one
that doens't).

Now there aren't separately named definitions (<number> and
<scientific-number>) but there are two definitions with the
same name.


Reverse some of action-2602

"some of", great. Also not linked to any minutes.
I can't find any reference to actions 2602 or 2615 in the wg mail archives. The first action on 12 June was 2616 and the last action given on 11 June was 2614.

So in summary we have

a) correct wording proposed by jwatt at a f2f and agreed by all present (including cameron)
b) a couple of mysterious actions which changed that wording, making it incompatible with what implementations do and with SVGT1.2

I am going to go through types.html and remove the erroneous "or in a
property's corresponding <a>presentation attribute</a>" throughout, and restore the good wording that jwatt proposed and we all agreed to.

Oh and about the EBNF? * means "zero or more" so 

number ::= integer ([Ee] integer)?
           | [+-]? [0-9]* "." [0-9]+ ([Ee] integer)?

already matches .3 and -.4 and .5e3 because there are zero [0-9].

I will add a test (on, say, stroke width) that checks that syntax if we don't already have one.

 Chris Lilley          
 Technical Director, Interaction Domain
 W3C Graphics Activity Lead
 Co-Chair, W3C Hypertext CG

Received on Wednesday, 5 May 2010 12:26:31 UTC