[boost] Generic Image Library (GIL) Review

From: Timothy M. Shead (tshead_at_[hidden])
Date: 2006-10-25 01:33:39

What is your evaluation of the design?
--------------------------------------

Excellent!  The design provides useful degrees of flexibility without
getting mired in generality-for-generality's-sake.

What is your evaluation of the documentation?
---------------------------------------------

Needs work - somehow the documentation in its present form manages to be
less than the sum of its parts.  Sorry for the vague hand-waving, but it
just seems as if the information you're looking for in the current
documentation is always Somewhere Else.  Having received very clear and
straightforward responses from the library authors on a number of
questions, I am confident that this is a matter of organization rather
than any fundamental flaw in design or content.  I would recommend
focusing on "howto" type documentation rather than the concepts and
reference docs.

What is your evaluation of the potential usefulness of the library?
-------------------------------------------------------------------

A must-have!  Images have become so ubiquitous in computer applications
that having a standard container for images makes as much sense as
standard containers for strings, vectors, etc.  

Did you try to use the library?  With what compiler?  Did you have any
problems?
----------------------------------------------------------------------

Yes - I used the library with gcc 3.4.4 on Gentoo Linux without any
difficulty.

How much effort did you put into your evaluation? A glance? A quick
reading? In-depth study?
-------------------------------------------------------------------

I retrofitted GIL into a mature 3D graphics pipeline, replacing the
home-grown floating-point HDR bitmap storage with GIL and using it to
implement several readers, filters (algorithms) and writers, and
integrating it with the gtkmm GUI toolkit for image previews.  Total
time spent with the library: ~ 5 hours.

Are you knowledgeable about the problem domain?
-----------------------------------------------

Yes, having implemented a bitmap storage library of my own that was also
parameterized on pixel type (albeit much more limited in scope).

Do you think the library should be accepted as a Boost library?
---------------------------------------------------------------

I recommend that GIL be accepted for inclusion in boost.

Regards,
Tim Shead