atom feed8 messages in org.oasis-open.lists.officeRe: [office] proposal; Amend fo:lette...
FromSent OnAttachments
Thomas ZanderJun 5, 2008 8:56 am.pgp
Jirka KosekJun 5, 2008 2:29 pm.pgp
Thomas ZanderJun 8, 2008 9:19 am.pgp
robe...@us.ibm.comJun 8, 2008 10:19 am 
Jirka KosekJun 8, 2008 1:58 pm.pgp
Thomas ZanderJun 8, 2008 2:35 pm.pgp
Michael Brauer - Sun Germany - ham02 - HamburgJun 9, 2008 4:13 am 
Thomas ZanderJun 9, 2008 6:17 am.pgp
Subject:Re: [office] proposal; Amend fo:letter-spacing
From:robe...@us.ibm.com (robe@us.ibm.com)
Date:Jun 8, 2008 10:19:53 am
List:org.oasis-open.lists.office

Thomas Zander <zan@kde.org> wrote on 06/05/2008 11:56:45 AM:

Hello TC,

In ODF the letter spacing attribute specifies the amount of space between letters. This is the way that XSL does things and this is fine for things like webpages. Unfortunately for DTP applications this is not really what is expected. The spacing between characters in DTP is not a static value, it is a percentage value. Basically when you have an 'm' the spacing adjustment is wider than when you have an 'i'. In KOffice we intend to use the percentage based letter spacing adjustment and we were surprised reading the spec (and trying out OOo) that this is not possible. Would it be possible to adjust the spec to also allow percentage based values?

Which would add the <ref name="percent"/> and we end up with the following;

<define name="style-text-properties-attlist" combine="interleave"> <optional> <attribute name="fo:letter-spacing"> <choice> <ref name="length"/> <value>normal</value> <ref name="percent"/> </choice> </attribute> </optional> </define>

I wonder if you already have the ability to do this with length? Specifically, if I set length to be 0.5em, does this really mean that exactly 0.5em is added after every character? Or does it mean that a nominal 0.5em is added, but the actual space applied is kerned according to typeface-specific ratios?

In other words, don't we have a model where we specify nominal spacing in the markup and then kerning occurs as an implementation-dependent algorithm?

-Rob