PI 1.8.0 RC6 Preview

[Juan Conejero]

Hi all,

I am now uploading installation packages for PixInsight 1.8.0 RC6 (commercial version only) to the Endor file server:

http://endor.uv.es/

They are on the "Forum Shared Files" repository, under the "1.8.0-RC6" folder. I've just uploaded a Mac OS X package, and I'm starting to build a Windows package right now.

I'd appreciate if you can test this new version, since I want to be sure it is fully stable on all platforms before making an official release. If you don't have an account on the Endor server, you can request one by writing us from our contact form, or directly to our support email address.

This version comes with important new features; some of them are:

- Integration of the SpiderMonkey 17 JavaScript engine (this is the same engine that was included in Firefox 17; about 10% faster, more stable and with more language features).

- Improved touch event support on image windows (track pads, touch screens):
* Pinch gestures: touch with two fingers on an image and pinch to zoom in/out.
* Pan gestures: press the spacebar key, touch and swipe with one finger to pan the view.

- Improved mouse wheel support on Mac OS X.

- Masked images are now highlighted with special view selector colors.

- Improved (completely rewritten, actually) Image core JavaScript object with image iterators (two times faster pixel access; I'll describe this on the development forum boards) and new methods (Image.multiscaleMedianTransform() is one of them).

- New versions of the ImageSolver, AnnotateImage and AperturePhotometry scripts by Andrés del Pozo. The solver script can now correct for arbitrary distortions, which makes it much more accurate than previous versions.

- Many bug fixes and stability improvements.

Thank you in advance.

[Josh Lake]

Fantastic! Do you think this is safe enough to try at NEAIC during my presentations? I'd love to show the folks the latest and greatest, but I also want it to be stable! RC5 has been working well on my OSX rMBP.

[Juan Conejero]

Hi Josh,

RC6 should be the most stable release of PixInsight ever. However, I have published this preview precisely to be 100% sure before making an official release. Please test the new application and tell me how it works. Let's see what other users have to say.

[Juan Conejero]

Installation packages for Linux and Windows are now available. Please test them out!

[Andres.Pozo]

Hi Juan,

I have found several warnings running my scripts that didn't happen in RC5. I will send you a new version by email.

Also I have found an error resetting the Javascript engine in Win7 x64. The test is very easy: you have only to start PI and execute the option "Reset JavaScript Runtime" from the menu EXECUTE in Script Editor.

[Juan Conejero]

Also I have found an error resetting the Javascript engine in Win7 x64. The test is very easy: you have only to start PI and execute the option "Reset JavaScript Runtime" from the menu EXECUTE in Script Editor.

This bug is now fixed. Thanks for detecting it.

It's either a bug in SM 17, or I am not understanding some changes in the way JavaScript contexts and runtimes have to be destroyed in this version (SM 17 introduces lots of critical changes with respect to previous versions). Fortunately, I have found a workaround so this won't happen anymore.

[Juan Conejero]

I have found several warnings running my scripts that didn't happen in RC5. I will send you a new version by email.

The new JavaScript compiler is better at detecting potential problems. It's also much stricter than previous versions. If you want to check your scripts against strict ECMA requirements, turn on the new strict compilation mode. Enter "help pjsr" on the console for more information.

[jtalbot]

Hi Juan,
Been playing with RC6 this morning at NEAIC and the program crashes each time I close an image.  Tried to attach my crash log but it has too many characters.  MAC OSX Mountain Lion.

Best,
Jon Talbot


   

[Josh Lake]

Here is my crash log from this morning, Juan.

[martin_magnan]

I also have experienced a few crash on window close but only after a long sequence of events. I havent been able to reproduce it.

The crash now occur when closing the window as opposed to when answering yes to the "Close this window anyway" dialog box.

PI 1.8 RC6 Preview on Mac OS X 10.8.3

Martin

[pfile]

Here is my crash log from this morning, Juan.

i guess it's good news that i have seen what seems to be the exact same crash. it was triggered during the loading of a project file.

[Philippe B.]

Hello
I played with RC6 on OSX 10.8.3 on MBP and I got no problem at all
I made several tests and seems very stable.

just a question
I put the console window in floating view and I closed it. Impossible to get it back, even by ALT+CMD+C or menu. Needed to reset settings to get it back 

Cheers
Philippe

[pfile]

i tried to integrate 250 bias frames on osx 10.8.2. i did not expect this to work because of the open file limits on osx, but i did not expect it to crash PI either... i think on 1.7 it would first fail to open one of the fits files and return to user control, but a second consecutive II run would then crash PI. here it just crashed straight away:

Code: [Select]

Thread 17 Crashed:
0   libsystem_kernel.dylib        0x00007fff8823c212 __pthread_kill + 10
1   libsystem_c.dylib              0x00007fff8f75caf4 pthread_kill + 90
2   libsystem_c.dylib              0x00007fff8f7a0dce abort + 143
3   QtCore                        0x000000010394e6e5 qt_message_output(QtMsgType, char const*) + 149
4   QtCore                        0x000000010394e8de 0x103935000 + 104670
5   QtCore                        0x000000010394eb1d qFatal(char const*, ...) + 141
6   QtCore                        0x0000000103a79071 QEventDispatcherUNIXPrivate::QEventDispatcherUNIXPrivate() + 2005
7   QtCore                        0x0000000103a7919c QEventDispatcherUNIX::QEventDispatcherUNIX(QObject*) + 40
8   QtCore                        0x000000010395898c 0x103935000 + 145804
9   libsystem_c.dylib              0x00007fff8f75b742 _pthread_start + 327
10  libsystem_c.dylib              0x00007fff8f748181 thread_start + 13

i believe that is the same place where 1.7 used to crash. hopefully this is easy to reproduce.

rob

[pfile]

after saving a project, and quitting PI, i got the 'printer crash':

Code: [Select]

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
0   libsystem_kernel.dylib        0x00007fff8823c212 __pthread_kill + 10
1   libsystem_c.dylib              0x00007fff8f75caf4 pthread_kill + 90
2   libsystem_c.dylib              0x00007fff8f7a0dce abort + 143
3   libc++abi.dylib                0x00007fff90ddda17 abort_message + 257
4   libc++abi.dylib                0x00007fff90ddb3c6 default_terminate() + 28
5   libobjc.A.dylib                0x00007fff89510887 _objc_terminate() + 111
6   libc++abi.dylib                0x00007fff90ddb3f5 safe_handler_caller(void (*)()) + 8
7   libc++abi.dylib                0x00007fff90ddb450 std::terminate() + 16
8   libc++abi.dylib                0x00007fff90ddc5b7 __cxa_throw + 111
9   com.pleiades-astrophoto.PixInsight 0x0000000100eb1ba7 pcl::CriticalSignalHandler(int) + 151
10  libsystem_c.dylib              0x00007fff8f7498ea _sigtramp + 26
11  ???                            0x00007fff7667bbe0 vtable for __cxxabiv1::__si_class_type_info + 16
12  com.apple.print.framework.PrintCore 0x00007fff8ddd8c37 PMGetUnadjustedPaperRect + 24
13  QtGui                          0x000000010161608c 0x1015d4000 + 270476
14  QtGui                          0x000000010161491e 0x1015d4000 + 264478
15  QtGui                          0x00000001016158f5 0x1015d4000 + 268533
16  QtGui                          0x00000001017ad66f QPrinter::paperSize() const + 35
17  com.pleiades-astrophoto.PixInsight 0x0000000100ad3311 pi::PixInsightPrinter::Save() const + 2213
18  com.pleiades-astrophoto.PixInsight 0x00000001000fba04 main + 18084
19  com.pleiades-astrophoto.PixInsight 0x0000000100005ce4 start + 52

strange that we are on this code path since i never tried to print anything... just quit PI.

[pfile]

hi, this is still on the RC6 private build, not the one released this morning. i'll have to try again.

this is a pretty bad crash. the IC tool gives a "Segmentation Violation" dialog box and then PI crashes without a coredump.

Code: [Select]

(gdb) where
#0  0x0000000100c6ad9e in pi::WorkspaceWindow::HideAlias ()
#1  0x0000000100037634 in pi::APIControlContext::SetControlUpdatesEnabled ()
#2  0x000000011bc8fae2 in pcl::SectionBar::__Click ()
#3  0x000000011bbbe85a in pcl::ControlEventDispatcher::MousePress ()
#4  0x000000010002d2b3 in pi::Control<QWidget>::mousePressEvent ()
#5  0x0000000101697d9c in QWidget::event ()
#6  0x000000010164d6b8 in QApplicationPrivate::notify_helper ()
#7  0x000000010164e008 in QApplication::notify ()
#8  0x0000000103a4b5e6 in QCoreApplication::notifyInternal ()
#9  0x0000000101655c28 in QApplicationPrivate::sendMouseEvent ()
#10 0x000000010160a301 in QMacCocoaAutoReleasePool::~QMacCocoaAutoReleasePool ()
#11 0x00007fff899a560e in -[NSWindow sendEvent:] ()
#12 0x0000000101601ab2 in QString::clear ()
#13 0x00007fff899a1744 in -[NSApplication sendEvent:] ()
#14 0x00000001016053c2 in QString::clear ()
#15 0x00007fff898b72fa in -[NSApplication run] ()
#16 0x000000010160e5f0 in operator!= ()
#17 0x0000000103a48a48 in QEventLoop::exec ()
#18 0x0000000103a4bd1b in QCoreApplication::exec ()
#19 0x00000001000fb96d in main ()

i discovered this quite by accident. one problem i have in 1.7 and now in 1.8 is that maybe 50% of the time, a file open dialog box will spontaneously close itself moments after it is opened. so i'll be trying to click on some file, but milliseconds before the window closes itself and i end up clicking on a UI element. in this case i double clicked on the "vv" pane hide button for the Output Filename section of ImageCalibration.

the steps to reproduce should be:

0. integrate a bunch of calibrated images (not sure if this is necessary, but it's what i did)
1. set up IC to calibrate some images; calibrate them with global apply (again, not sure if necessary)
2. when finished just start clicking on the hide/show buttons on different parts of the IC pane. should crash eventually.

rob