<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#ffffff" text="#000000"> Cyrille Berger a écrit : <blockquote cite="mid:201001211713.52492.cberger@cberger.net" type="cite"> <pre wrap="">On Thursday 21 January 2010, Olivier BERTEN wrote: </pre> <blockquote type="cite"> <pre wrap="">There's a mix of different reasons, the least not being my lack of knowledge of XML and programming skills... I first went from <RGB r="1" g="1" b="1"> to <RGB values="1 1 1"> simply because it was easier to deal with different color models the same way (you were talking about laziness ;-) ) </pre> </blockquote> <pre wrap="">Well, I don't know if I am spoiled by the framework I use for xml, but for me accessing an attribute of an element is as simple as "red = element.attribute("r",0.0).toDouble()" :) The two reasons we went for that form is that in "1 1 1" the order of channels is not specified, and it is also not possible to validate it with a schema (there have been changed on XML schemas to also validate data, but I don't know if it would work in that case). </pre> </blockquote> I understand the validation point but I'm not really sure of the interest in validating data... since <sRGB r="20" g="-100" b="0.2"/> would validate but it wouldn't make any sense... but it might be my bad knowledge of xml ;-) For the code question, mine is "item.values[(elem.attrib['model'],elem.attrib['space'])] = map(eval,elem.text.split())". So it's the same code for any model, which makes sense since I don't display the colors at parsing time but later in the process, where I use data provided in another "format" anyway. <blockquote cite="mid:201001211713.52492.cberger@cberger.net" type="cite"> <pre wrap=""></pre> <blockquote type="cite"> <pre wrap="">For the rest (metadata, swatch organization) wich aren't in the create draft, I made some choices but I'm already reconsidering most of them... </pre> </blockquote> <pre wrap=""> What are the choice ? I think we would aim to use things like RDF/XMP, but neither of them are "simples", so I don't know if we should not also have a simple metadata spec that we can reuse (something based on dublin core, maybe). </pre> </blockquote> My main choice was to not use (yet) any existing spec because it would complexify my xml and at this early developement stage I want to keep it easy to read for humans. But this is most probably going to change. Olivier </body> </html>