Mac test build based on Qt5.11.1 - please test

[Dirk Hohndel]

Hi

If you have access to a Mac (running macOS 10.11 or newer - 10.10 is no longer supported), I'd appreciate if you could test the latest binary that I have created.
This still crashes with the user manual, so Qt5.11.1 wasn't the magic fix for that. But this gets us ready to experiment with the QML compiler...

http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-48-g03bc40194d26.dmg

This binary was created on a different Mac - I'm in the process of shifting the infrastructure that I use to build Mac and iOS binaries, so I want to make double certain that this gets well tested.

I hope I'll have time this weekend to update Travis to also build macOS against the same Qt5.11.1 and then work on migrating iOS to that version as well (including Travis).

Thanks for your help

/D

[Federico Masias]

Hey Dirk,

Quick test:

- Printing appears blurry for any dive profile other than the one dive
template. It's been awhile since I've touched this, so I'm not sure if this
is new to this version or goes further back
- Connecting to facebook causes a crash. Attached is a log from that
crash. Tested a few times, from different screens and same result.
- User manual does indeed still crash.

Let me know if there's anything specific you'd like tested. I'm on macOS
10.13, but unfortunately, I have no access to BT/BLE dive computers. I'm
realizing I have to be up in a couple of hours so I'll check back later
today (tomorrow?)

- Federico

Post by Dirk Hohndel
Hi
If you have access to a Mac (running macOS 10.11 or newer - 10.10 is no
longer supported), I'd appreciate if you could test the latest binary that
I have created.
This still crashes with the user manual, so Qt5.11.1 wasn't the magic fix
for that. But this gets us ready to experiment with the QML compiler...
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.
0-48-g03bc40194d26.dmg
This binary was created on a different Mac - I'm in the process of
shifting the infrastructure that I use to build Mac and iOS binaries, so I
want to make double certain that this gets well tested.
I hope I'll have time this weekend to update Travis to also build macOS
against the same Qt5.11.1 and then work on migrating iOS to that version as
well (including Travis).
Thanks for your help
/D
_______________________________________________
subsurface mailing list
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

[Benjamin]

Subsurface starts up and gives me a red bar at the bottom of the screen
saying "Unmatched action 'hash' ", but that could really be my setup, given
that Dirk just saved me from a bad data file :) (Thank you, Dirk)

Running using Subsurface -v -v
When trying to import from my Petrel 2, I choose "classic mode" and get the
following messages on the CLI:

Starting download from BT

Starting the thread 0

IOBluetooth works only on the main thread or a thread with a running
CFRunLoop

Failed to connect to device 00:13:43:0C:56:29 . Device state
QBluetoothSocket::UnconnectedState
. Error: QBluetoothSocket::SocketError( -2 )

qt_ble_open( 00:13:43:0C:56:29 )

failed to connect to the controller 00:13:43:0C:56:29 with error "Remote
device cannot be found"

Finishing download thread: "Unable to open 00:13:43:0C:56:29 Shearwater
(Petrel 2)"

If I choose "BLE mode", then I get the following messages on the CLI:

Starting download from BT

Starting the thread 0

qt_ble_open( 00:13:43:0C:56:29 )

failed to connect to the controller 00:13:43:0C:56:29 with error "Remote
device cannot be found"

Finishing download thread: "Unable to open LE:00:13:43:0C:56:29 Shearwater
(Petrel 2)"

In both cases, if I try to save the libdivecomputer logfile, I simply get a
single line in it:

Subsurface: v4.8.0, built with libdivecomputer v0.7.0-devel-Subsurface-NG
(02560a7e7fe82919d584d3edbf3876f90382052c)

What can I do to try and give more information form my side?

Benjamin

Post by Federico Masias
Hey Dirk,
- Printing appears blurry for any dive profile other than the one dive
template. It's been awhile since I've touched this, so I'm not sure if this
is new to this version or goes further back
- Connecting to facebook causes a crash. Attached is a log from that
crash. Tested a few times, from different screens and same result.
- User manual does indeed still crash.
Let me know if there's anything specific you'd like tested. I'm on macOS
10.13, but unfortunately, I have no access to BT/BLE dive computers. I'm
realizing I have to be up in a couple of hours so I'll check back later
today (tomorrow?)
- Federico

Post by Dirk Hohndel
Hi
If you have access to a Mac (running macOS 10.11 or newer - 10.10 is no
longer supported), I'd appreciate if you could test the latest binary that
I have created.
This still crashes with the user manual, so Qt5.11.1 wasn't the magic fix
for that. But this gets us ready to experiment with the QML compiler...
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-48-g03bc40194d26.dmg
This binary was created on a different Mac - I'm in the process of
shifting the infrastructure that I use to build Mac and iOS binaries, so I
want to make double certain that this gets well tested.
I hope I'll have time this weekend to update Travis to also build macOS
against the same Qt5.11.1 and then work on migrating iOS to that version as
well (including Travis).
Thanks for your help
/D
_______________________________________________
subsurface mailing list
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

_______________________________________________
subsurface mailing list
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

[Dirk Hohndel]

Subsurface starts up and gives me a red bar at the bottom of the screen saying "Unmatched action 'hash' ", but that could really be my setup, given that Dirk just saved me from a bad data file :) (Thank you, Dirk)

No, I see the same thing here. I'm looking into this.

Running using Subsurface -v -v
Starting download from BT
Starting the thread 0
IOBluetooth works only on the main thread or a thread with a running CFRunLoop
Failed to connect to device 00:13:43:0C:56:29 . Device state QBluetoothSocket::UnconnectedState . Error: QBluetoothSocket::SocketError( -2 )
qt_ble_open( 00:13:43:0C:56:29 )
failed to connect to the controller 00:13:43:0C:56:29 with error "Remote device cannot be found"
Finishing download thread: "Unable to open 00:13:43:0C:56:29 Shearwater (Petrel 2)"

That's interesting. The message on IOBluetooth is a new one - I have noticed that Qt5.11 gives better error messages in a few other situations as well. Also something to investigate.
I assume that you are able to download from your Petrel 2 with the release version of Subsurface 4.8?
Which version of macOS are you on? I created an issue on GitHub to track this: https://github.com/Subsurface-divelog/subsurface/issues/1472 <https://github.com/Subsurface-divelog/subsurface/issues/1472>

Starting download from BT
Starting the thread 0
qt_ble_open( 00:13:43:0C:56:29 )
failed to connect to the controller 00:13:43:0C:56:29 with error "Remote device cannot be found"
Finishing download thread: "Unable to open LE:00:13:43:0C:56:29 Shearwater (Petrel 2)"

Has BLE to your Petrel 2 worked with 4.8?

Subsurface: v4.8.0, built with libdivecomputer v0.7.0-devel-Subsurface-NG (02560a7e7fe82919d584d3edbf3876f90382052c)

That makes sense - Subsurface doesn't manage to open the connection, so libdivecomputer never gets to play.
I do notice that this says v4.8.0 - so was this even written by the 4.8.0.48 binary that you tried? Or do we write an incorrect version number?

Thanks for testing. This really helps figure out what we need to fix!

/D

[Dirk Hohndel]

Hi Benjamin,

Here's an update...

Post by Dirk Hohndel

Subsurface starts up and gives me a red bar at the bottom of the screen saying "Unmatched action 'hash' ", but that could really be my setup, given that Dirk just saved me from a bad data file :) (Thank you, Dirk)

No, I see the same thing here. I'm looking into this.

This is fixed in master.

Post by Dirk Hohndel

Running using Subsurface -v -v
Starting download from BT
Starting the thread 0
IOBluetooth works only on the main thread or a thread with a running CFRunLoop
Failed to connect to device 00:13:43:0C:56:29 . Device state QBluetoothSocket::UnconnectedState . Error: QBluetoothSocket::SocketError( -2 )
qt_ble_open( 00:13:43:0C:56:29 )
failed to connect to the controller 00:13:43:0C:56:29 with error "Remote device cannot be found"
Finishing download thread: "Unable to open 00:13:43:0C:56:29 Shearwater (Petrel 2)"

That's interesting. The message on IOBluetooth is a new one - I have noticed that Qt5.11 gives better error messages in a few other situations as well. Also something to investigate.
I assume that you are able to download from your Petrel 2 with the release version of Subsurface 4.8?
Which version of macOS are you on? I created an issue on GitHub to track this: https://github.com/Subsurface-divelog/subsurface/issues/1472 <https://github.com/Subsurface-divelog/subsurface/issues/1472>

If you look at this closely you see that it first fails BT because the device is in UnconnectedState and then it tries BLE and that fails as well.
Now, the BLE part of the Petrel 2 has been failing for us for a while now.
I played here with the BT download for a little bit and this is my result:

If I pair my Petrel (Linus currently has my Petrel 2, so I tested with the BT-only Petrel) with my Mac with system tools,
and then click on '...' in order to open the BT dialog and search for the dive computer, select the Petrel and then Save it,
and then click Download,
wait for macOS to pop up another pairing request, enter '0000' as pin and OK,
then it still shows the same IOBluetooth error, but it downloads successfully.
And I can show that same behavior with both stock Subsurface 4.8.0 (Qt5.10.1) and with the test binary I created based on Qt5.11.1.

Can you reproduce the successful download, following these steps?

/D

[Dirk Hohndel]

Post by Federico Masias
Hey Dirk,
Printing appears blurry for any dive profile other than the one dive template. It's been awhile since I've touched this, so I'm not sure if this is new to this version or goes further back

Could you compare with 4.8.0 (the release version)? I'll look into that here as well.

Post by Federico Masias
Connecting to facebook causes a crash. Attached is a log from that crash. Tested a few times, from different screens and same result.

I created an issue for this: https://github.com/Subsurface-divelog/subsurface/issues/147 <https://github.com/Subsurface-divelog/subsurface/issues/147>1

Post by Federico Masias
User manual does indeed still crash.

https://github.com/Subsurface-divelog/subsurface/issues/14 <https://github.com/Subsurface-divelog/subsurface/issues/14>61

Post by Federico Masias
Let me know if there's anything specific you'd like tested. I'm on macOS 10.13, but unfortunately, I have no access to BT/BLE dive computers. I'm realizing I have to be up in a couple of hours so I'll check back later today (tomorrow?)

I really appreciate the quick test and that you found things that I usually don't test. Any additional testing you could do of other features (basically going through the menus and playing around) in order to uncover other hidden problems... (of course this applies to everyone here...)

Thanks again

/D

[Dirk Hohndel]

Post by Federico Masias
Connecting to facebook causes a crash. Attached is a log from that crash. Tested a few times, from different screens and same result.

I created an issue for this: https://github.com/Subsurface-divelog/subsurface/issues/1471 <https://github.com/Subsurface-divelog/subsurface/issues/1471>

For those not following along on GitHub, it turns out Subsurface 4.8.0 on Mac built with Qt5.10.1 already crashes with Facebook. The fact that we haven't received a user bug report might indicate how rarely this feature is used... :-(

/D

[Federico Masias]

Dirk,

The 4.8 release version appears to be the same. It's been quite some time
since the last time I looked at printing, but I could've sworn that at some
point the graphs from the same computer appeared smoother.

In any case, I've attached PDFs of what I'm seeing when I print with each.
It's not nearly as bad as I thought it was at 2am.

- Federico

Post by Federico Masias
Hey Dirk,
- Printing appears blurry for any dive profile other than the one dive
template. It's been awhile since I've touched this, so I'm not sure if this
is new to this version or goes further back
Could you compare with 4.8.0 (the release version)? I'll look into that here as well.
- Connecting to facebook causes a crash. Attached is a log from that
crash. Tested a few times, from different screens and same result.
I created an issue for this: https://github.com/
Subsurface-divelog/subsurface/issues/1471
- User manual does indeed still crash.
https://github.com/Subsurface-divelog/subsurface/issues/1461
Let me know if there's anything specific you'd like tested. I'm on macOS
10.13, but unfortunately, I have no access to BT/BLE dive computers. I'm
realizing I have to be up in a couple of hours so I'll check back later
today (tomorrow?)
I really appreciate the quick test and that you found things that I
usually don't test. Any additional testing you could do of other features
(basically going through the menus and playing around) in order to uncover
other hidden problems... (of course this applies to everyone here...)
Thanks again
/D

[Murillo Bernardes]

Post by Dirk Hohndel
Hi
If you have access to a Mac (running macOS 10.11 or newer - 10.10 is no longer supported), I'd appreciate if you could test the latest binary that I have created.
This still crashes with the user manual, so Qt5.11.1 wasn't the magic fix for that. But this gets us ready to experiment with the QML compiler...
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-48-g03bc40194d26.dmg

Works fine for the basic things and download correctly from Perdix AI.

Breaks on user manual and Facebook as already reported.

[Dirk Hohndel]

Post by Murillo Bernardes

Post by Dirk Hohndel
Hi
If you have access to a Mac (running macOS 10.11 or newer - 10.10 is no longer supported), I'd appreciate if you could test the latest binary that I have created.
This still crashes with the user manual, so Qt5.11.1 wasn't the magic fix for that. But this gets us ready to experiment with the QML compiler...
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-48-g03bc40194d26.dmg

Works fine for the basic things and download correctly from Perdix AI.
Breaks on user manual and Facebook as already reported.

Thanks. I created a GitHub issue for the Facebook crash. https://github.com/Subsurface-divelog/subsurface/issues/1471
And we are already tracking the user manual issue there as #1461

/D

[Dirk Hohndel]

OK, going back to the beginning of the thread with an update...

It turns out (and I am VERY happy about that) that the Facebook and User Manual crashes were caused by mis-compilation of QtWebKit. With the help of the maintainer I got that sorted and I think I now have a binary that works again.

http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-66-g12519af8b434.dmg

I'd really appreciate if the Mac owners could once again give this a good test. Please especially test things that we don't spend as much time with - dive computer configuration, FB, user manual, filtering, etc. I haven't forgotten about the question regarding print quality - it's on my TODO list, but I first wanted to get the crashes sorted out.

If this looks stable, I'll aim for a 4.8.1 fairly quickly...

/D

Post by Dirk Hohndel
Hi
If you have access to a Mac (running macOS 10.11 or newer - 10.10 is no longer supported), I'd appreciate if you could test the latest binary that I have created.
This still crashes with the user manual, so Qt5.11.1 wasn't the magic fix for that. But this gets us ready to experiment with the QML compiler...
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-48-g03bc40194d26.dmg
This binary was created on a different Mac - I'm in the process of shifting the infrastructure that I use to build Mac and iOS binaries, so I want to make double certain that this gets well tested.
I hope I'll have time this weekend to update Travis to also build macOS against the same Qt5.11.1 and then work on migrating iOS to that version as well (including Travis).
Thanks for your help
/D
_______________________________________________
subsurface mailing list
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

[Murillo Bernardes]

Post by Dirk Hohndel
OK, going back to the beginning of the thread with an update...
It turns out (and I am VERY happy about that) that the Facebook and User Manual crashes were caused by mis-compilation of QtWebKit. With the help of the maintainer I got that sorted and I think I now have a binary that works again.

Great news.

Post by Dirk Hohndel
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-66-g12519af8b434.dmg
I'd really appreciate if the Mac owners could once again give this a good test. Please especially test things that we don't spend as much time with - dive computer configuration, FB, user manual, filtering, etc. I haven't forgotten about the question regarding print quality - it's on my TODO list, but I first wanted to get the crashes sorted out.

- Facebook works, post correctly.
- User manual works, looks good.
- Filtering works correctly, although the "clear" icons seems weird,
but looks like this happens on previous versions as well.
- Downloads correctly from Shearwater Petrel AI (BLE) and Oceanic OCi (serial).
- I don't have a computer that can be configured, so not tested.

[Dirk Hohndel]

Murillo,

Post by Murillo Bernardes

Post by Dirk Hohndel
http://subsurface-divelog.org/downloads/test/Subsurface-4.8.0-66-g12519af8b434.dmg
I'd really appreciate if the Mac owners could once again give this a good test. Please especially test things that we don't spend as much time with - dive computer configuration, FB, user manual, filtering, etc. I haven't forgotten about the question regarding print quality - it's on my TODO list, but I first wanted to get the crashes sorted out.

- Facebook works, post correctly.
- User manual works, looks good.
- Filtering works correctly, although the "clear" icons seems weird,
but looks like this happens on previous versions as well.
- Downloads correctly from Shearwater Petrel AI (BLE) and Oceanic OCi (serial).
- I don't have a computer that can be configured, so not tested.

Thank you so much for that detailed test. That's really appreciated.
I noticed an oddity in the way I build the binaries, so I'm making some changes to the build setup and will have another test binary in a little while, but this is all very encouraging.

/D

[Dirk Hohndel]

Hi Bryan

Hey, I'm new to the mailing list (signed up today!) and to using Subsurface (I am however a dev, so figured I'd just jump in head-first when I found this project).

Yay! We love seeing new users and even more love seeing new developers!

Log => Add Dive => Photos => Load Image(s) from File(s)
- then I select an image (jpg) from file explorer, and hit enter
Then the app crashes.

So this is with the binary I referenced below, not with something you built from source, correct?

Didn't CC the list, wasn't sure if I should? Mainly wasn't sure if this was a real problem or if I'm just a supernoob to SS and screwed something up.

Yes, please always Cc the list when you think you find bugs or problems. I took the liberty of simply adding the list here.
And pleeeeaaaase don't call the project SS - that's the Schutz Staffel of Nazi Germany and some of us (like, err, me) really don't like that abbreviation used for Subsurface... If you really have to use an abbreviation, ssrf might work. But it's not that many letters, is it?

Let me know if I should go digging for logs or if you were able to replicate. Can also make a mini-video.

Typically when an app crashes on the Mac it offers to report the bug. Take it up on that offer and look at the stack traces that it shows - especially for the thread that crashed. Then send that stack trace with your report. Or dig in and fix the bug :-)

In this case I wasn't able to reproduce your crash here - but that might only mean that I didn't quite hit the scenario that you used in the first place. Can you give more information, maybe even send a test image plus the precise steps to reproduce this?

Thanks so much

And again: WELCOME

/D

[Bryan Oliver]

Thanks! Sorry about the abbreviation, didn't even think about that but
yikes, very good point.

Yes it was with your referenced binary, think I had a typo in my first
email ("built" instead of "build"). I'll pull the crash log when I get
home, and make a gif as well.

Thanks again!
B

Post by Dirk Hohndel
Hi Bryan

Hey, I'm new to the mailing list (signed up today!) and to using

Subsurface (I am however a dev, so figured I'd just jump in head-first when
I found this project).
Yay! We love seeing new users and even more love seeing new developers!

I noticed when I ran the QT 5.11.1 built on MBP (High Sierra 10.13.4)
Log => Add Dive => Photos => Load Image(s) from File(s)
- then I select an image (jpg) from file explorer, and hit enter
Then the app crashes.

So this is with the binary I referenced below, not with something you
built from source, correct?

Didn't CC the list, wasn't sure if I should? Mainly wasn't sure if this

was a real problem or if I'm just a supernoob to SS and screwed something
up.
Yes, please always Cc the list when you think you find bugs or problems. I
took the liberty of simply adding the list here.
And pleeeeaaaase don't call the project SS - that's the Schutz Staffel of
Nazi Germany and some of us (like, err, me) really don't like that
abbreviation used for Subsurface... If you really have to use an
abbreviation, ssrf might work. But it's not that many letters, is it?

Let me know if I should go digging for logs or if you were able to

replicate. Can also make a mini-video.
Typically when an app crashes on the Mac it offers to report the bug. Take
it up on that offer and look at the stack traces that it shows - especially
for the thread that crashed. Then send that stack trace with your report.
Or dig in and fix the bug :-)
In this case I wasn't able to reproduce your crash here - but that might
only mean that I didn't quite hit the scenario that you used in the first
place. Can you give more information, maybe even send a test image plus the
precise steps to reproduce this?
Thanks so much
And again: WELCOME
/D

[Berthold Stoeger]

Post by Bryan Oliver
Thanks! Sorry about the abbreviation, didn't even think about that but
yikes, very good point.
Yes it was with your referenced binary, think I had a typo in my first
email ("built" instead of "build"). I'll pull the crash log when I get
home, and make a gif as well.

This is ultimately related to #1445: https://github.com/Subsurface-divelog/
subsurface/issues/1445

The add picture-dialog will access the wrong dive. If there isn't any dive
yet, it will access a non-existing dive and crash.

desktop-widgets/simplewidgets.cpp, l.383:
QDateTime time_first = QDateTime::fromTime_t(first_selected_dive()->when,
Qt::UTC);

There is no "first_selected_dive()" -> kaboom.

Berthold