Re: A few issues with thumbs.el

[Mathias Dahl]

Mathias Dahl <address@hidden> writes:


>> The above-mentioned specification is well thought out, and the list
>> of advantages of this approach is convincing.  So the best thing for
>> us to do now is just to follow this standard and to share the same
>> standard thumbnail directory with other GNU programs.
>
> Point taken. I will put this on my todo-list for tumme.el. It should
> not be that hard to implement.

I did some investigations yesterday and found that there is something
I'm not sure how we could solve in a good way in Emacs: the tEXt
chunks (text annotations) in PNG files.

The ImageMagick package contains a utility called `identify' that can
read these values in the following way:

$ identify -verbose my_file.png | grep Thumb::

So, as both tumme.el and thumbs.el already use ImageMagick, although
only the convert and mogrify tools (tumme.el could use only convert
too if I wanted it too, but mogrify simplifies things a bit), I see no
problem in how to read those annotations.

The problem is writing them. After searching the net for a while I
have found two ways to do it, using external tools:

* pngcrush - a command line tool for compressing PNG to make them
smaller, but which also can add tEXt chunks to PNG files.

* a small perl-script, png-tEXt.pl, which sole purpose is to add text
  annotations to PNG files.

I'm not sure about the licence for them, but they seem free enough (at
least gratis) for use to use them.

What I do not like is involving yet another utility and I have
therefore a question:

Would it be possible to (and would we want to, Emacs is after all
mostly a text-processing tool, not a image manipulation app), in the
long run, learn Emacs to read and write these annotations? If I
understand it correctly pngcrush uses libpng, and that is what Emacs
uses too, right?

The subject of thumbnails seems to be sensitive, so I'll concentrate
on fixing other things in tumme.el, in the meantime... :)