[CREATE] OpenRaster reference library - libora

[Luka Čehovin]

> All of krita is written in C++ and using QtXML, I don't think anything can be
> used.

Ok. no problem. Some performance improving ideas are still welcome (I
have found out that writing zip files is slow even when removing all
the python wrappers and intermediate files that are used in mypaint
... not that slow, but slower than expected. I will experiment with
different settings ...).

>
>> >> For now my plans are to add thumbnail support and global metadata (are
>> >> there any specifications on that?) ...
>> >
>> > No, not yet. I think we are aiming at support for RDF/XMP, with something
>> > like that:
>> >
>> > <layer ... metadata="data/layer1metadata.xmp" />
>> > and
>> > <image ... metadata="data/imagemetadata.rdf" />
>> >
>> > Unless we want to have more than one metadata file per-layer, and do
>> > something like:
>> > <layer ...>
>> >  <metadata type="xmp" file="data/layer1metadata.xmp" />
>> >  <metadata type="rdf" file="data/layer1metadata.rdf" />
>> > </layer>
>>
>> This sounds like it is still being discussed. And I think this is the
>> main problem of the standard or the proposal ... almost nothing is
>> fixed.
> What is in the drafts is fixed.

Ok. I have also checked the discussion page and found out that many
things are still being discussed (not much recent activity though). I
will therefore really try to focus on the things that look fixed and
try to make them stable.

>
>> I really have little interest in implementing something that
>> does not have some consensus and formalization ... one can hardly call
>> it a standard then. What I am trying to propose from the beginning
>> (and nobody wants to comment apparently) is that some simple version
>> of the standard is formalized and everything that is still discussed
>> is postponed. We insert some version identifier in the format and call
>> the current standard version 1.0 for example.
>
> I would be fined with that.

if ("fined" == "fine") then: we should then separate the part that is
well defined to a separate wiki page. I like the idea of having
multiple specifications with different levels of complexity. Basically
the main thing what we need at the moment is some kind of standard
identifier that would mark the type/version of the document. Something
similar to mimetype.

>
>> Then we start from there
>> and make things more complex. As the most evolved applications (like
>> Gimp and Krita) use their own formats I see no problem with that ...
>
> One of the goal for us, in Krita, is to replace our own format with
> OpenRaster.

that is great!

-- 
Luka Čehovin
http://luka.tnode.com