[ImageJ-devel] Please don't quit

Wed Nov 5 13:59:25 CST 2014

Hi Lee,

> So CellProfiler is all caught up, at least it's at 2.0.0 rc 16 -

Awesome!

> hopefully largely locked-down from an API perspective.

Yeah, the API is pretty solid at this point -- especially SciJava Common,
which is thoroughly out of beta. The only backwards-incompatible change
planned is a rework of org.scijava.io, but I hope that won't impact most
downstream code.

The other big eventual changes on the ImageJ2 side will be unification of
the net.imglib2.meta and net.imagej data structures, as well as
net.imglib2.ops and net.imagej.ops APIs. If you are a heavy user of
imagej-common or imagej-ops, you may see some disruption there in a few
months' time.

In general, if you see breakages and disruption while keeping things up to
date, you're welcome to complain. Especially if the migration path is not
obvious.

We are going through some related pain on the ImgLib2 front right now, with
imglib2 coming out of beta a couple weeks ago. We still haven't sent the
release announcement because it is not uploaded to the ImageJ update site
yet -- we have to first upgrade all affected Fiji components to the new
version. But after this point the imglib2 API will be extremely stable, so
ultimately it's a good thing.

Regards,
Curtis

On Wed, Nov 5, 2014 at 1:50 PM, Lee Kamentsky <leek at broadinstitute.org>
wrote:

> Great Curtis, I even made it say "ImageJ for CellProfiler" in the system
> info. So CellProfiler is all caught up, at least it's at 2.0.0 rc 16 -
> hopefully largely locked-down from an API perspective.
>
> On Wed, Nov 5, 2014 at 12:33 PM, Curtis Rueden <ctrueden at wisc.edu> wrote:
>
>> Hi Lee,
>>
>> > Please don't quit
>>
>> I would suggest avoiding a custom service. There are two other (hopefully
>> simpler!) ways you can achieve your goal -- see below.
>>
>> First, a summary of the architecture:
>>
>> The current DefaultAppService responds to AppQuitEvents as follows:
>>
>> protected void onEvent(@SuppressWarnings("unused") final AppQuitEvent
>> event) {
>> getApp().quit();
>> }
>>
>> Which delegates the operation to the App plugin with the highest priority.
>>
>> The following App plugins ship with net.imagej:imagej:
>>
>> * org.scijava.app.SciJavaApp [LOW priority] -- bundled with
>> org.scijava:scijava-common
>> * io.scif.SCIFIOApp [NORMAL priority] -- bundled with io.scif:scifio
>> * net.imagej.legacy.LegacyImageJApp [NORMAL priority] - bundled with
>> net.imagej:imagej-legacy
>> * net.imagej.app.ImageJApp [HIGH priority] -- bundled with
>> net.imagej:imagej-common
>> * net.imagej.app.ToplevelImageJApp [HIGH+1 priority] -- bundled with
>> net.imagej:imagej
>>
>> So:
>> * If net.imagej:imagej is on the classpath, the ToplevelImageJApp will
>> take precedence.
>> Otherwise:
>> * If net.imagej:imagej-common is on the classpath, the ImageJApp takes
>> precedence.
>>
>> However, ToplevelImageJApp extends ImageJApp and does not override
>> quit(). So either way, the quit behavior will be the same:
>>
>>   public void quit() {
>>     commandService.run(QuitProgram.class, true);
>>   }
>>
>>
>> https://github.com/imagej/imagej-common/blob/imagej-common-0.10.1/src/main/java/net/imagej/app/QuitProgram.java#L79-L96
>>
>> Note that the devastating System.exit(0) is only called when
>> exitWhenQuitting is set to true. This is part of OptionsMisc:
>>
>>
>> https://github.com/imagej/imagej-common/blob/imagej-common-0.10.1/src/main/java/net/imagej/options/OptionsMisc.java#L75-L76
>>
>> So, there are two potential ways I would suggest to achieve your goal:
>>
>> 1) Set the OptionsMisc exitWhenQuitting to false. This is handy if you
>> don't need a lot of control over exactly what happens when quit() is
>> called, but you just want to stop the System.exit(0) from firing.
>>
>> 2) Create a CellProfilerApp plugin and set it to a higher priority than
>> ImageJ. This solution makes the most sense to me because it gives you more
>> flexibility over the behavior of a couple other operations, too: about()
>> and prefs().
>>
>>
>> https://github.com/scijava/scijava-common/blob/scijava-common-2.35.0/src/main/java/org/scijava/app/App.java
>>
>> Regards,
>> Curtis
>>
>> On Wed, Nov 5, 2014 at 10:23 AM, Lee Kamentsky <leek at broadinstitute.org>
>> wrote:
>>
>>> Oops never mind, what I tried actually worked. AppQuitEvent.consume()
>>> did the trick. Go figure - you guys really threw the whole kitchen sink
>>> into it, didn't you?
>>>
>>> @Plugin(type = Service.class, priority = Priority.HIGH_PRIORITY)
>>> public class CellProfilerAppService extends DefaultAppService {
>>> static boolean canQuit = false;
>>> public static void allowQuit() {
>>> canQuit = true;
>>> }
>>> public static void preventQuit() {
>>> canQuit = false;
>>> }
>>> @EventHandler
>>> public void onEvent(final AppQuitEvent event) {
>>> if (canQuit) {
>>> super.onEvent(event);
>>> } else {
>>> final UIService uiService = getContext().getService(UIService.class);
>>> final LogService logService = getContext().getService(LogService.class);
>>> if (uiService.isVisible()) {
>>> UserInterface ui = uiService.getDefaultUI();
>>> logService.info("Quit action: hide the application frame");
>>> ui.getApplicationFrame().setVisible(false);
>>> } else {
>>> logService.info("Quit action: do nothing");
>>> }
>>> event.consume();
>>> }
>>> }
>>>
>>> }
>>>
>>>
>>> On Wed, Nov 5, 2014 at 11:15 AM, Lee Kamentsky <leek at broadinstitute.org>
>>> wrote:
>>>
>>>> Hi all,
>>>> I had a class (
>>>> https://github.com/CellProfiler/CellProfiler/blob/master/java/src/main/java/org/cellprofiler/ijutils/CellProfilerAppEventService.java)
>>>> that implemented the deprecated AppEventService whose purpose was to
>>>> prevent ImageJ from quitting if a user closed its window by overriding
>>>> AppEventService.quit(). Quitting is pretty devastating for CellProfiler
>>>> since the process closes when the user's probable intent was to hide the
>>>> window.
>>>>
>>>> I'm hoping someone can give me a hint about how to do it now - I
>>>> thought I'd cheat by asking instead of figuring it out myself.
>>>>
>>>> Thanks in advance,
>>>> --Lee
>>>>
>>>
>>>
>>> _______________________________________________
>>> ImageJ-devel mailing list
>>> ImageJ-devel at imagej.net
>>> http://imagej.net/mailman/listinfo/imagej-devel
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://imagej.net/pipermail/imagej-devel/attachments/20141105/3792a3f6/attachment.html>