[ImageJ-devel] Auto Threshold in ImageJ2

[Curtis Rueden]

Hi Gabriel,

> > Can you clarify what is the reason not to be doing under GPL2 or 3?
>
> We do not get funding by the NIH if we insist on GPL, hence the move
> to the BSD license.

In addition to funding, it is also about the ImageJ community itself, which
is quite diverse. There are many people (e.g., Bob Dougherty) who use
ImageJ who rely on the fact that it is not copyleft-licensed. By using GPL
as the license for ImageJ2 core, we would be splintering the community,
which is something we are very much trying to avoid doing. There will of
course continue to be a place for GPL-licensed code within the Fiji project.

> > Would you be willing to relicense that code to BSD?
>
> Yes, that is fine.
> Same for the Auto Local Threshold if you are planning to make it a
> built in command.

Thank you!

> > It would piss me off if somebody decides to make a derivative code
> > which is closed source. But perhaps you can clarify this a bit?
>
> In reality, those people who make derivative closed source code are
> just in for a lot of pain. They will get no support, take all the
> blame for bugs, do not get any updates automatically, and in general
> are despised. Not to mention ridiculed by sending a mail to them, Cc:
> the public mailing lists, pointing to Curtis' brilliant Open Science
> text.
>
> So I am not worried about such people.

It is true that by licensing the code under BSD, people will have the
ability to incorporate it into closed-source products. But as Johannes
says, there are substantial tradeoffs to doing that.

> Now that we are at it, I haven't checked the histogram in the IJ2
> applet, but in IJ1, the frequency of the largest bin gets trimmed down
> (for display purposes) when it is too large to still be able to show
> what the frequencies in the other bins are. While this might be useful
> for display (so you do not see only one large bin value), it is quite
> confusing if it is not clearly explained beforehand (there is a
> mismatch between the listed histogram and the displayed one). Can this
> be an option, rather than the only way it works?

Thanks for the suggestion. Barry is currently making good progress on
threshold support in ImageJ2, but we haven't implemented histograms yet. We
will certainly avoid that sort of silent magic—I completely agree that
there should be an explicit toggle for that sort of thing.

Regards,
Curtis


On Wed, Feb 6, 2013 at 2:41 AM, Gabriel Landini <G.Landini at bham.ac.uk>wrote:

> On Monday 04 Feb 2013 22:50:37 Johannes Schindelin wrote:
> > we would like to include your code (or a derivative thereof) in ImageJ2.
> > Just to be sure: the code in question is this one:
>
> Hi Dscho,
>
> >
> http://fiji.sc/cgi-bin/gitweb.cgi?p=fiji.git;a=blob;f=src-plugins/Auto_Thres
> >
> hold/src/main/java/fiji/threshold/Auto_Threshold.java;h=d09e56d5671c91d367f3
> > 9607375bb0b40290f7e4;hb=refs/heads/master
> >
> > Would you be willing to relicense that code to BSD?
>
> Yes, that is fine.
> Same for the Auto Local Threshold if you are planning to make it a built in
> command.
>
> Now that we are at it, I haven't checked the histogram in the IJ2 applet,
> but
> in IJ1, the frequency of the largest bin gets trimmed down (for display
> purposes) when it is too large to still be able to show what the
> frequencies
> in the other bins are. While this might be useful for display (so you do
> not
> see only one large bin value), it is quite confusing if it is not clearly
> explained beforehand (there is a mismatch between the listed histogram and
> the
> displayed one). Can this be an option, rather than the only way it works?
>
> Cheers
>
> Gabriel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://imagej.net/pipermail/imagej-devel/attachments/20130207/2964d67f/attachment.html>