Fits Header integration time

You get the right EXPTIME = 50 sec
I really don’t understand why I get EXPTIME = 10 sec
Here is the result of a new pre-processing with Siril 0.99.10
Screenshot 2021-06-16 at 11.00.38.pdf (655.1 KB)

Are you working on a Mac ? Maybe the bug is only in the macOS version .

I don’t see why there would be a difference between linux (which is UNIX) and macOS in this part.

Please, share the whole result.fits file.

Here it is : http://dl.free.fr/ipShWtOij

@rbarbera we need your help :slight_smile:
Can you reproduce it with the version you compile on mac?

If yes I will need you output some debug line and I will tell you where.

Hi, sorry for the delay. On first place: the official 0.99.10 (the one that was published last week) is also failing on my machine with my own dataset. Then I’ve checkedout the current main branch and compiled it locally. This version didn’t fails and performs as expected. I’ve stacked 40x240s (9600s total)

I’m adding evidences of this behavior:

Individual frame

Stack with the official version:

Strack with my own build:

Obviously right now, I’ve not changed anything on the source code.

Regards

Wow thanks, that’s a funny one!

Probably a story about lib version …

Well, in fact they are different (at the minor number, but different)

On my computer, I have 2.68.2

brew list glib
/usr/local/Cellar/glib/2.68.2/lib/libgio-2.0.0.dylib
/usr/local/Cellar/glib/2.68.2/lib/libglib-2.0.0.dylib
/usr/local/Cellar/glib/2.68.2/lib/libgmodule-2.0.0.dylib
/usr/local/Cellar/glib/2.68.2/lib/libgobject-2.0.0.dylib
/usr/local/Cellar/glib/2.68.2/lib/libgthread-2.0.0.dylib

Inside the package you are using 2.64.2

Inside the bundle
   2,5M 10 jun 10:40 libgio-2.0.0.dylib
   1,6M 10 jun 10:40 libglib-2.0.0.dylib
    39K 10 jun 10:40 libgmodule-2.0.0.dylib
   431K 10 jun 10:40 libgobject-2.0.0.dylib

Looking inside the strings, we can use the symbols reference to know the version
➜  lib strings libgio-2.0.0.dylib  | grep 2
....
G_IS_APP_INFO (appinfo2)
../../../../gtk/source/glib-2.64.2/gio/gbufferedinputstream.c
../../../../gtk/source/glib-2.64.2/gio/gbufferedoutputstream.c
../../../../gtk/source/glib-2.64.2/gio/gbytesicon.c
../../../../gtk/source/glib-2.64.2/gio/gcancellable.c
../../../../gtk/source/glib-2.64.2/gio/gcharsetconverter.c

I can reproduce with our AppImage. This is linked with the issue 723.

I don’t see how a library change would impact the simple task of adding exposure times?

That depends of g_date_time probably. Will investigate I can almost reproduce the AppImage

@vinvin @rbarbera @JCJ : I think I found it.
Not an easy one …

Here: GDateTime: GLib Reference Manual

You can see there is the %f, and I use it. However, looks like this was introduced in the glib 2.66 version. But the documentation says nothing about it.

1 Like

Bug is now probably fixed in master.

Good catch!

Well done effectively !

Issue solved in 0.99.10-1.
Thanks to the team !

3 Likes