Hi Klaus,

On 09/01/2011 10:45 AM, Klaus Drechsler wrote:

However, I tstill think that something is wrong.

1. This registering/unregistering happens only to "PlanarFigureObjectFactory" (as you can see from the attachment in the last post). After all this "PlanarFigureObjectFactory" registering/unregistering output the application starts lightning fast. Should there not be similar output for all other plugins/modules? (I compiled *everything* in, so I would expect that)
If I remember it correctly, the PlanarFigure module is the only one which does the object factory registration/unregistration automatically at static initializtation and de-initialization time.

2. This slow startup and multiple registering/unregistering does not only happen once after compilation (as "release" btw), but at every startup. So, the "re-caching" and checking of time stamps seems not to work as expected.
Yes, this is a bug ;-) Recently, colleagues of mine also started complaining...

At least on Windows... there was a problem in CTK with the saved time-stamp accuracy. This is fixed since today and with MITK commit 0db796 the fixed CTK sources are used. So please update your MITK sources and do a build at superbuild level.


3. Why is there a difference in output if I start "AwesomeApp" (or an external app created using the generator) or the "extapp" delivered with mitk? (both take long to start at *every* startup, but only the former shows this "PlanarFigureObjectFactory" stuff at the beginning).
I don't know exactly... the set of libraries loaded is not completely the same, and maybe the PlanarFigure module cannot be unloaded in these cases.

The most important question however is:

How can I (temporarily) disable this caching mechanism (or whatever is responsible for the slow startup)? I currently see no advantage for us having this feature and it is a bit annoying during development :)

You definitely do not want to disable the caching ;-) it would make things even slower. Anyway, with the bug fix the startup time should be okay, hopefully.

If I can provide more information and/or if you can give me a hint at which places I should have a look at, please let me now.


Thanks for insisting that something is wrong,

Sascha

Best wishes,
 Klaus


----- Ursprüngliche Nachricht -----

Von: Sascha Zelzer

Gesendet: 31.08.11 23:32 Uhr

An: Klaus Drechsler

Betreff: Re: [mitk-users] Multiple Registering/Unregistering of PlanarFigureObjectFactory during Startup


Hi Klaus,

This is not a bug.

During the first application start-up, the CTK Plugin Framework loads and unloads each installed plug-in (and hence all its dependencies) to cache the resources compiled into them. Unfortunately, this is awefully slow on Windows (especially in Debug mode).

After the first start, the framework keeps track of the mofified time stamps of the plug-isn and re-caches the resources only in case the plug-in was modified since the last start.

Best,
Sascha

On 08/31/2011 12:09 PM, Klaus Drechsler wrote:
Hi,

I am encountering a problem during startup of an external app created using the plugin generator. Startup takes unusually long and in the dos box I see several time the registration and unregistration of PlanarFigureObjectFactory (see attached output.txt).

This happened since I moved to the ctk-based MITK (hash: 32fff1f68c797f392c2b98c7ae35766dc079819f ).

Is this a known bug? Is there a quick´n dirty fix for it?


Best wishes,
 Klaus

PS: Startup also takes longer when I just start "extapp" and I hear that the hardisk is *very* busy. However, here I do not see the registration/unregistration of PlanarFigureObjectFactory.


------------------------------------------------------------------------------
Using storage to extend the benefits of virtualization and iSCSI
Virtualization increases hardware utilization and delivers a new level of
agility. Learn what those decisions are and how to modernize your storage 
and backup environments for virtualization.
http://www.accelacomm.com/jaw/sfnl/114/51434361/
_______________________________________________
mitk-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mitk-users

Reply via email to