[Image-SIG] Pillow: the "unfriendly" fork?

Gora Mohanty gora at mimirtech.com
Thu Dec 20 20:56:05 CET 2012

On 21 December 2012 00:30, Alex Clark <aclark at aclark.net> wrote:
> Folks,
> I'm trying to mentally prepare for adding Python 3 support to Pillow before
> PyCon 2013. To do so, I plan to review the work that has been done here:
> - https://github.com/fluggo/Pillow
> Then merge and release.
> Until this point, we've been attempting to track changes via "upstream"
> tickets in the hope that one day there'd be a new PIL release and we'd stop
> Pillow development. However I now suspect that the more likely scenario is
> that we will have to declare at some point that Pillow is "the unfriendly
> PIL fork", meaning that we will not continue to track changes in Pillow
> along with the "upstream" changes in PIL.
> Does this sound reasonable? I think the "best" time to do this would be with
> the addition of Python 3 support to Pillow, along with a corresponding
> significant release number change e.g. "Pillow 2.0.0 (now with Python 3
> support)"

I hesitate to comment, not having contributed to PIL in
any way other than having been a happy user, and still
using it from time to time. I also absolutely do *not* mean
to point any fingers, as I understand how difficult it can
be to support a widely-used FOSS project, and I do appreciate
the key role that PIL has played, and am grateful for the time
that people have spent on it on a voluntary basis.

Nevertheless, here are some points:
1. We had a client willing to pay to add support for certain
    TIFF images that were not then supported by PIL. There
    was one response from an independent developer to a
    post on the list offering to pay for someone to add such
    support. Being able to add such support directly to PIL
    would have been a big plus, but for various reasons we
    decided to switch to Imagemagick.
2. Imagemagick seems to have a larger community, more
    recipes on the Internet, and a larger feature set (please
    correct me if I am wrong).
3. The Python support for Imagemagick is still pretty bad,
    but at least for us the pros outweigh the cons. So much
    so, that we are currently willing to use it via subprocess
    calls to convert/montage/etc.

So, a revamped PIL, in whatever form, would be great. If
you are driving the effort, I would say go ahead, by all means.
However, what I would seriously consider getting involved in
is a project for real Python support for Imagemagick.


More information about the Image-SIG mailing list