[Bookreader] Annotations, revisited

[Andreas Gros]

On 20.01.2010 23:24, Sayamindu Dasgupta wrote:
> On Thu, Jan 14, 2010 at 11:04 AM, Samuel Klein<meta.sj at gmail.com>  wrote:
>> I just chatted this afternoon with Andi Gros, who is quite interested
>> in seeing annotations added to our bookreader's repertoire.  I hope we
>> can make some progress there, now that we have a growing active
>> audience of users, who will want more than author and title to figure
>> out what texts to use for what purposes!
>>
>> Starting with something simple like "a global/local-network list of
>> reviews/notes/comments organized by title" without worrying about
>> different versions of works or annotations of specific words or
>> sentences, could be very usable.
>>
>> Is anyone actively working on something like this atm?  I also chatted
>> briefly with the founder of RatePoint, who have been following OLPC
>> for some time and interested in making global reviews by 'product' or
>> 'entity' happen with a simple web interface.
>>
>
> A major pain point is supporting annotations in a file-format
> independent way. For example, currently, the PDF standard has pretty
> extensive support for annotations, but that is of very little value in
> Read as we try to deal with other formats as well. With EPUB, I can
> have really flexible annotations, or even editable books (the
> underlining support as shown in
> http://sayamindu.randomink.org/images/read_highlight.gif actually
> makes the EPUB editable for a while), but all of this does not make
> any sense in a PDF context.
I agree, IMO it is important to -- at least initially -- treat 
annotations in a way that is as much as possible independent from the 
actual underlying format which is to be annotated. As PDF and EPUB, and 
possibly any other format, offer different editing possibilities, to me 
the easiest way would be to not use either, but to come up with 
something independent. Sounds mad? Probably, but lets assume that even 
if a PDF does not contain text but images, it displays the content in 
single pages -- no matter how large these actually are -- and tells the 
viewer (e-reader, ....) how big (width X and height Y) the page is. And 
lets assume that text-highlighting (to me these are not really 
annotations and depend too strongly on the actual display of text on a 
page/in an image) is to be treated differently from textual annotations. 
I am referring to the latter ones below.

For texts/textual works, annotations can be made to
     (a) a text/work
     (b) a page
     (c) a coordinate on a page / a rectangular area on a page (two 
coordinates: upper left corner + lower right corner)

Options (a) and (b) don't depend in any way on the display of the single 
page, but could be treated independently and could/should be solved as a 
first step.

Option (c) is more tricky. If we store the x/y coordinates of the 
annotation relative to the page width X and height Y, it is up to the 
viewer/e-reader to interpret that and assuming that the reader knows 
what X and Y are, this should be possible.
Please correct me if I'm wrong :-)

My approach would be to start with solving options (a) and (b) and once 
we get into the flow, consider solving (c).

What do you think?


> As for indexing/storage of annotations, I have tried to keep the
> options open by storing the notes and bookmarks in a single, global
> sqlite database, but it can become complicated as things like editable
> books are enabled.
Could you please point me to where you implemented the "note taking"? I 
haven't found that in the code of the read activity that I checked out 
with git (my bad). That sounds as if we could use that as a first entry 
point to solving (a) and (b). :-)

Thank you very much!

Andi