Migrate MATLAB scripts #19
Conversation
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Sometimes, you only want to use parts of Fiji as a library, not start the interactive ImageJ. To do this, start 'Miji(false)'. By default, ImageJ will still be started (e.g. 'Miji' does the same as 'Miji(true)'). Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
While at it, almost rewrite the thing to avoid unnecessary cd'ing. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jean-Yves Tinevez <jean-yves.tinevez@pasteur.fr>
Signed-off-by: Jean-Yves Tinevez <jeanyves.tinevez@gmail.com>
Signed-off-by: Jean-Yves Tinevez <jeanyves.tinevez@gmail.com>
Unfortunately, a restart is required if Java3D was not installed, but it is as convenient now as we could make it, at least. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This file only contains comments, as it is meant to be published using Matlab's publish() function. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jean-Yves Tinevez <jeanyves.tinevez@gmail.com>
Signed-off-by: Stephan Preibisch <stephan.preibisch@gmx.de>
We also have to ensure that the bin folder in the MATLAB jre path is writable, otherwise it will confuse the user. Reported by Peter Beemiller Signed-off-by: Jean-Yves Tinevez <jeanyves.tinevez@gmail.com>
Now that Johannes fixed the surface plot mode of the 3D viewer, we can put this demo back. It is a demo explainging how to get the MATLAB logo rendered in 3D in Fiji 3D viewer. Signed-off-by: Jean-Yves Tinevez <jean-yves.tinevez@pasteur.fr>
Apparently, the light was bluish, dunno whether we can do that, too :O) Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jean-Yves Tinevez <jeanyves.tinevez@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
... also if the user asked the main window to be hidden... Noticed and reported by Walter de Back. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
In Miji.m, there it a call to fiji.User_Plugins.installScripts(); Unfortunately, this causes a nasty bug with MATLAB: calling this static method modifies the static MATLAB java path, which is normally forbidden. The consequences of that are nasty: adding a class to the dynamic class path can be refused, because it would be falsy recorded in the static path. On top of that, the static path is fsck in a weird way, with file separator from Unix, causing a mess on Windows platform. So we give it up as now. Signed-off-by: Jean-Yves Tinevez <jeanyves.tinevez@gmail.com>
In the same vein as the previous commit... Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Interoperability for the win. Signed-off-by: Jean-Yves Tinevez <jean-yves.tinevez@pasteur.fr>
…rrayImg Signed-off-by: Jean-Yves Tinevez <jean-yves.tinevez@pasteur.fr>
The Miji.m script adds Fiji.app/jars and Fiji.app/plugins to the MATLAB classpath by explicitly iterating over *.jar in these directories. Thus it was not recursing to subdirectories - notably /jars/bio-formats. The script will now recurse through subdirectories to find jars. Adapted from: http://stackoverflow.com/questions/2652630/ how-to-get-all-files-under-a-specific-directory-in-matlab
Instead of manually adding filesep let's let MATLAB decide how to make a full file path.
This proposed change removes a bug I described on imagej-dev . Any other use of javaaddpath, say to develop one's other software, loses the instance of MIJ. This is because javaaddpath automatically calls clear java. By returning the MIJ instance as a workspace object, the MIJ instance will not be cleared. The only change for the user is to call: MIJ = Miji; instead of just Miji; Signed-off by: Mark Hiner
This changes all relevant scripts from javax.vecmath to org.scijava.vecmath, since we now ship vecmath as a standard dependency!
Java 3D 1.6 now ships with Fiji. It does not need to be (and cannot be!) installed separately.
…pt spam of printing every artifact added to the classpath to the console. Related commit: 2633d12 - Remove `cd ..` to stay in the same directory while loading Miji - Only turn warnings off for: `MATLAB:javaclasspath:jarAlreadySpecified`
|
Cool! In the ImageJ side, it may be called scripts, but in MATLAB terminology, many of these files are MATLAB functions rather than scripts. How about a folder named Also how about adding a folder called |
bug fix for permute
ImageJ.m src\scripts\update
My reason for putting the files in |
|
I now included the changes from #20 (#22) and #21 as well. @kouichi-c-nakamura let me know if any of your changes are missing in this branch, as I merged https://github.com/kouichi-c-nakamura/imagej-matlab/tree/ImageJ_split which hopefully reflects #20. |
|
Thank for merging. With regard to Do you guys agree that just adding Java class paths is equivalent of having a headless ImageJ instance? Or is it equivalent of If YES, then I'll get rid of If NO, then I'll change the As to the folder for MATLAB functions to be used from within MATLAB, I suggest to use Another folder to think about is for MATLAB test code. They can live in |
|
Unfortunately, I continue to be too busy with other priorities to maintain imagej-matlab properly. However, @kouichi-c-nakamura and @imagejan, if you are both satisfied with the state of this PR, I will merge it as-is; let me know. In good news, I will be working with @yuminguw to bring him up to speed on this codebase, so that he can start lending a hand with imagej-matlab support and maintenance. It will take some time, but hopefully we can get the project moving again relatively soon. |
|
Better late than never... |
I migrated the history of
scripts/*.minfiji/fijito this repository, and then moved these scripts intosrc/main/resources/script_templates.Should we consider any other location for these scripts?
//@kouichi-c-nakamura