[ome-devel] HPC/OMERO integration

[Johannes Schindelin]

Hi Alex,

I keep you in Cc: since I do not assume you are subscribed to ome-devel.
However, it is asking too much to reinstate the complete Cc: list of me.

In the future, you might consider this: not everybody is subscribed to
every list, so if you cull the Cc: list it is like sending a mail to the
wrong recipient.

On Mon, 5 Mar 2012, Alex Herbert wrote:

> Thanks for the help with the ImageJ headless situation. I would like to
> try ImageJ2 if you think it will support BioFormats from within
> command-line macros.

ImageJ2 does not support macros yet other than through the ImageJ1 legacy
layer.

> I decided to go down the route of building my own headless.jar having
> read about the solution used in Fiji. I found some old code for the
> headless modifications but it was based on an old version of ImageJ and
> changes to ImageJ since meant that I could not run my plugin. I cannot
> even remember what did not work but having seen how easy it was to
> modify the code I just did it myself using the latest stable release of
> the relevant ImageJ classes. My headless system now works fine and I
> have complete source control of the changes so can fully support my
> environment.

The downside is that we have now diverging code, and it will cost much
more work to reconcile.

> I did not know about the new Javassist workaround. I will have to try
> running Fiji with my command-line macros and see if it works OK. This
> will mean that I wouldn't have to maintain my headless code.

As soon as I knew about your intentions, I added documentation.

> My unsolved issue is not being able to use BioFormats. Thanks for adding
> some extra insight to this problem. Having looked at the code (and used
> the plugin) I know it modifies the GenericDialog. But I did not persist
> with my investigations as I now load my images using the native TIFF
> support in ImageJ (this was a path of least resistance). It would be
> nice to use BioFormats to read OME-TIFFs from OMERO and avoid having to
> do extra work with lots of single plane TIFFs.

This is one of the very negative outcomes of diverging code: you have to
solve a problem again that we already solved. Most likely, there are also
problems that you have solved and we have not.

> I would be happy to try using ImageJ2 as the engine for my server
> scripts if it works with BioFormats support. However I have failed to
> find documentation on ImageJ2 and running it with some options like -h,
> -help, --help did not give any clues. If you could provide me with an
> example for the command-line mode I will try it out.

The modifier in the package name of ImageJ2 states that it is an Alpha
version. As such, it should not be surprising that it does not support any
command line option yet.

For the moment, the best way to run ImageJ2 in a server process is to
control it using Java, or even better, to write plugins for it in Java.

The design of the ImageJ1 macro language is so fundamentally not able to
support server processes that I would strongly recommend not to use macros
in this context. Yes, you might get lucky with some macros. But that seems
not to have been the case here.

Ciao,
Johannes