Mike, I've had second thoughts about the current proposal:
<0x1F><image record id><3-byte width+height><1-byte depth><1-byte alt tag len>
Mainly, this breaks the format specification. Older viewers won't
find and display these images. We really shouldn't switch formats
without updating the version number. It's a BAD thing; it's Wrong.
Secondly, this doesn't extend to images with associated larger
versions.
Alternatively, and my choice, would be to find a backwards-compatible
way of doing this. I can think of two ideas.
The first is that we use the function code as a "hint" preceding a
regular image function code (or big-image function code). We'd use
<0x1e><2-byte width><2-byte height><1-byte depth><1-byte alignment>
meaning "remember these; they're for the next image in the stream".
Unfortunately, this won't work for the alt tag; a viewer that doesn't
understand this function code won't be able to skip the text. Note
that I've added alignment; I'd really like to be able to keep the
"align={left|right|center}" information from the IMG element.
The second is to store this information in the metadata record instead
of directly in a text record. We'd add a new metadata subrecord,
which would contain a list of image-record info blocks. Each info
block would contain the record-id, width, height, depth, alt-tag (if
any), and any other interesting image info we can think of. This has
the advantage of being able to store the ALT info. It has the
disadvantage of making the info a bit more tedious to get to.
Comments?
Bill
_______________________________________________
plucker-dev mailing list
[EMAIL PROTECTED]
http://lists.rubberchicken.org/mailman/listinfo/plucker-dev
