[epub-cfi] side bias

[Daniel Weck]

The side bias is expressed with either 'b' (before) or 'a' (after). I
suggest using characters other than 'a' and 'b' for the HTML text, to
avoid confusing CFI examples such as:
/1:3[bbb;s=b]
Also note that the HTML examples have 'a' strings coming before 'b'
strings, which is the exact opposite than 'b'=>before and 'a'=>after
for the side bias notation ! :) One more reason to replace the strings
with "xxx<br/>yyy" or such like.

Additionally, maybe it is just me, but I would personally prefer
"before" first, and then "after" in this sentence:

"
It can take two values: a means that the location is glued to the
content after it and b to the content before it.
"

Because 'a' comes before 'b' in the alphabet, I first thought that 'a'
meant "left" and 'b' meant "right" :) I then realized the connection
with "after" and "before".

Cheers, Dan

[Garth Conboy]

Certainly no objection to tuning examples to use characters other and a's and b's in sample text -- yes, good suggestion.

As far as changing the string used to express before and after, from "b" and "a" to their longer spellings, I'll let others comment -- no strong feelings.

Best,
Garth

[Daniel Weck]

On 5 May 2011, at 22:02, Garth Conboy wrote:
> As far as changing the string used to express before and after, from
> "b" and "a" to their longer spellings, I'll let others comment -- no
> strong feelings.

Note that I did not suggest that :) (just to be sure)
In fact, I wouldn't vote for s=before or s=after...it's too verbose.

Going back to the sentence starting with "It can take two values:" ==>
"a" comes before "b", which is somewhat illogical because of their
corresponding "after" and "before" meanings. To remove the potential
ambiguity resulting from alphabetical order between 'a' and 'b', maybe
we could use the binary '0' and '1' terms for side bias values.

It's a little thing, but I bet some readers will get caught at first
read.

Dan

[matt.g...@bell.net]

I don't think we can win on the identifiers. Granted these are anglo-centric, but 1 and 0 are equally confusing as they are better as booleans than describing sidedness (I think interpreting as the "a" side coming before the "b" side might be a common first thought for these, though).
 
I'll look at fixing the examples to use some other character data as that will improve the spec readability, but will leave a/b alone for now (at least).
 
> Subject: Re: [epub-cfi] side bias
> From: ga...@google.com
> Date: Thu, 5 May 2011 14:02:43 -0700
> To: epub-work...@googlegroups.com

[Daniel Weck]

Hi, my concerns have been addressed in the latest draft. Thanks! Dan

http://epub-revision.googlecode.com/svn/trunk/build/linking/cfi/epub-cfi.html#sec-path-side-bias