PixInsight Prerelease Version: 1.8.4.1190

[Juan Conejero]

Here is a new prerelease version of PixInsight: 1.8.4.1190. The Linux and Windows versions are now being uploaded to software distribution. The OS X versions are now available on a temporary folder of our website:

For commercial licenses:
https://pixinsight.com/tmp/1.8.4/PI-macosx-x86_64-01.08.04.1190-20151112-c.zip

For trial licenses:
https://pixinsight.com/tmp/1.8.4/PI-macosx-x86_64-01.08.04.1190-20151112-t.zip

This OS X version has full support for high-dpi modes, making use of the native physical resolution of Retina displays. See this forum post for more information with an example.

Unfortunately, there is still a weird bug in the OS X version, which is completely innocuous but very ugly. This bug causes image windows to be rendered as flat gray panels occasionally, especially after closing modal dialogs. The problem gets fixed by just zooming in/out the view in question, but things like these don't give a very good impression, especially to new users, and definitely don't make justice to the overall quality of the platform. So I have to fix this problem before the official release. This is why I haven't uploaded these packages to software distribution.

Let me know how it goes. Hopefully we'll have an official release next week. Enjoy!

[Juan Conejero]

Just a note to let you know that the Linux and Windows versions are now available on software distribution. Thank you.

[mschuster]

Hi Juan,

Thank you. My images look really excellent at your 1:2 "actual size" on iMac 5k with your improvements!

Maybe it is a bit confusing to need to use 1:2 to get an "actual size" view? I understand why, but it doesn't seem quite natural. But it sure looks good!

Thanks,
Mike

[TechnoPhil]

I am trying it on my MacBook Pro Retina 15".
Stable but some icons are not "Retina ready". 

[pfile]

- updater3 seems to have not crashed
- my project loads without segfault
- switching workspaces works well

- can't test high-DPI, don't have any retina macs

hey! transparent windows during drag! glad to see that this finally made it to OSX!

rob

[Sean]

Hi Juan,

Thanks for all the work on the OS X version. I'm about to buy a new iMac 5k and am looking forward to using the new version of PixInsight.

Sean

[Juan Conejero]

Thank you everybody!

- my project loads without segfault

Glad to know that. However, there is still a bug that will be fixed in the official release. When you have the Statistics window open with track view active, if you select a view by clicking it with the mouse, sometimes this segfaults on all platforms. The same may happen with HistogramTransformation, but is much more unlikely. This is a small problem that I have already identified. It will be fixed next Monday.

[pfile]

ok thanks for the heads-up, i have not been able to use it too much yet, but i am trying right now.

one thing i just noticed is that even though "use native file IO dialogs" is checked in the global prefs, when i select Execute Script File... i get a Qt-style dialog. i don't think it used to be that way.

thanks,

rob

[cdesselles]

I'm not too keen on pre-releases.  Not enough IT savvy.  So when may we expect the full commercial release? (approximately)

[Philippe B.]

I tested on Win64 and OSX 10.11 (El Capitan).
[/size]This is fine.
[/size]On OSX I have no discover any problem (for the moment    )
[/size]
[/size]Thank you
[/size]

[pfile]

i am seeing a problem where sometimes PI hangs when writing out a FITS file. it might be related to dismissing the warning that FITS can not support an STF... seems like if i dismiss that too fast that it sometimes hangs, but not sure.

i have to kill PI after this happens... the pause/abort button is greyed out. some info from lldb:

Code: [Select]

(lldb) thread list
Process 39959 stopped
* thread #1: tid = 0x24f74ae, 0x00007fff8a6644de libsystem_kernel.dylib`mach_msg_trap + 10, queue = 'com.apple.main-thread', stop reason = signal SIGSTOP
  thread #2: tid = 0x24f74cd, 0x00007fff8a66a232 libsystem_kernel.dylib`kevent64 + 10, queue = 'com.apple.libdispatch-manager'
  thread #3: tid = 0x24f74e6, 0x00007fff8a66948a libsystem_kernel.dylib`__semwait_signal + 10, name = 'pi::PixInsightWatchDog'
  thread #4: tid = 0x24f7501, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'WebCore: IconDatabase'
  thread #5: tid = 0x24f77ae, 0x00007fff8a6644de libsystem_kernel.dylib`mach_msg_trap + 10
  thread #6: tid = 0x24f77c2, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'QFileInfoGatherer'
  thread #7: tid = 0x24f77c4, 0x00007fff8a6693fa libsystem_kernel.dylib`__select + 10, name = 'Qt bearer thread'
  thread #8: tid = 0x24f77c7, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::BlockFree'
  thread #9: tid = 0x24f77c8, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::Marking'
  thread #10: tid = 0x24f77c9, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::Marking'
  thread #11: tid = 0x24f77ca, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::Marking'
  thread #12: tid = 0x24f77cb, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::Marking'
  thread #13: tid = 0x24f77cc, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::Marking'
  thread #14: tid = 0x24f77cd, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'JavaScriptCore::Marking'

not sure if thread 6 is the relevant thread or not, but here is where it's stopped:

Code: [Select]

(lldb) thread select 6
* thread #6: tid = 0x24f77c2, 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10, name = 'QFileInfoGatherer'
    frame #0: 0x00007fff8a669136 libsystem_kernel.dylib`__psynch_cvwait + 10
libsystem_kernel.dylib`__psynch_cvwait:
->  0x7fff8a669136 <+10>: jae    0x7fff8a669140            ; <+20>
    0x7fff8a669138 <+12>: movq   %rax, %rdi
    0x7fff8a66913b <+15>: jmp    0x7fff8a664c53            ; cerror_nocancel
    0x7fff8a669140 <+20>: retq   

forcing this thread to return (a few times) had no effect and eventually i had to kill PI and start over.

rob

[mschuster]

Hi Rob,

The workaround here is to wait until the autohide process console window open/close completes, and then click to dismiss the fits/xisf dialog. As you say, don't click too fast, or you risk losing your work. This problem is being worked on.

Thanks,
Mike

[Juan Conejero]

I can reproduce this problem consistently on OS X by quickly pressing Enter to dismiss dialogs. Another obscure Qt bug on OS X, and a very tough one... I'm trying to figure out a workaround right now.

[TechnoPhil]

Be careful because I noticed that if PI crashes, many big temp files into /tmp remains undeleted!

[Juan Conejero]

Be careful because I noticed that if PI crashes, many big temp files into /tmp remains undeleted!

That depends on what you were doing before the crash. I can't be careful during a crash... or there would be no crash

All files in /tmp are deleted automatically by the operating system with the default OS X configuration.