Re: Subsurface-mobile status

2017-04-04 Thread Dirk Hohndel 
On Tue, Apr 04, 2017 at 09:52:05PM +0200, Marco Martin wrote:
> On Tuesday 04 April 2017, Dirk Hohndel wrote:
> > Thanks for catching this and pointing it out. Something we need to fix
> > before pushing this out as a public beta, of course.
> > 
> > I'd love if you have patches - my next two days will be very busy, not
> > sure I'll find much time to look into this.
> 
> seen what i did wrong, now on my master it's ok.
> I've also changed the look of those boxes to be on top so they don't get 
> under 
> the virtual keyboard anymore

I played with this. The edit seems to work now, based on some very
preliminary testing. But the scaling of the profile is off.

I'll drop that commit for now until I understand better what's happening
there, but push the rest and create a new APK so others can test.

http://subsurface-divelog.org/downloads/daily/Subsurface-mobile-4.6.3.200-arm.apk

/D

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

Re: Shearwater Perdix test result, was Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Linus Torvalds 
On Tue, Apr 4, 2017 at 2:34 PM, Lutz Vieweg  wrote:
>
> But whenever I tried to start the download, the operation failed after a
> wait time of ~5 seconds with a dialog stating:
>>
>> Insufficient privileges to open the device D4:60:3A:**:**:** Shearwater
>> (Perdix)
>
> "strace" did not reveal any "ENOPERM"-failing operations or such.

That error message is actively misleading for BT devices. It's just
"open failed".

It's from the serial line (and USB device) case where the most common
reason for being unable to open a device was due to permission issues
on the device node.

So what happens is that dc_device_open() returns a failure (_any_
failure), and we fail to "stat()" the device name. The stat() will
obviously always fail for bluetooth, since there isn't a device node
at all.

We should probably try to add some "report_error()" logic to the
actual libdivecomputer code so that there would be more information
about exactly what actually failed.

 Linus
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Shearwater Perdix test result, was Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Lutz Vieweg 

As promised earlier today, I just had a small window of opportunity this evening
to check whether downloading from a friend's just-bought Shearwater Perdix
(firmware release v44) with Subsurface under Linux works.

Hardware tested on:
- Desktop computer (AMD Ryzen 7 1800X, Asus X370 Pro)
- Bluetooth adapter: 0a12:0001 Cambridge Silicon Radio, Ltd Bluetooth Dongle
  (Bluetooth 4 LE capable, according to its specs)

Software:
- Arch Linux, linux-4.10.6 kernel
- bluez 5.44-1, bluez-utils 5.44
- subsurface 4.6.3-5

Pairing looked just fine: (serial numbers obfuscated with "*")


bluetoothctl

[NEW] Controller 00:1A:7D:**:**:** ryzen [default]
# agent KeyboardOnly
Agent registered
# default-agent
Default agent request successful
# scan on
Discovery started
[NEW] Device D4:60:3A:**:**:** Perdix
# info D4:60:3A:**:**:**
Device D4:60:3A:**:**:**
Name: Perdix
Alias: Perdix
Paired: no
Trusted: no
Blocked: no
Connected: no
LegacyPairing: no
# pair D4:60:3A:**:**:**
Attempting to pair with D4:60:3A:**:**:**
Pairing successful
# connect D4:60:3A:**:**:**
Attempting to connect to D4:60:3A:**:**:**
Connection successful
[CHG] Device D4:60:3A:**:**:** Connected: yes
[NEW] Primary Service
/org/bluez/hci0/dev_D4_60_3A_**_**_**/service000a
1801--1000-8000-00805f**
Generic Attribute Profile
[NEW] Primary Service
/org/bluez/hci0/dev_D4_60_3A_**_**_**/service000b
fe25c237-0ece-443c-b0aa-e02033**
Vendor specific
[NEW] Characteristic
/org/bluez/hci0/dev_D4_60_3A_**_**_**/service000b/char000c
27b7570b-359e-45a3-91bb-cf7e70**
Vendor specific
[NEW] Descriptor
/org/bluez/hci0/dev_D4_60_3A_**_**_**/service000b/char000c/desc000e
2902--1000-8000-00805f**
Client Characteristic Configuration
[NEW] Descriptor
/org/bluez/hci0/dev_D4_60_3A_**_**_**/service000b/char000c/desc000f
2901--1000-8000-00805f**
Characteristic User Description
[CHG] Device D4:60:3A:**:**:** UUIDs: 1800--1000-8000-00805**
[CHG] Device D4:60:3A:**:**:** UUIDs: 1801--1000-8000-00805**
[CHG] Device D4:60:3A:**:**:** UUIDs: fe25c237-0ece-443c-b0aa-e0203**
[CHG] Device D4:60:3A:**:**:** ServicesResolved: yes
[CHG] Device D4:60:3A:**:**:** Paired: yes
# trust D4:60:3A:**:**:**
Changing D4:60:3A:**:**:** trust succeeded
[CHG] Device D4:60:3A:**:**:** Trusted: yes
# info D4:60:3A:**:**:**
Device D4:60:3A:**:**:**
Name: Perdix
Alias: Perdix
Paired: yes
Trusted: yes
Blocked: no
Connected: yes
LegacyPairing: no
UUID: Generic Access Profile(1800--1000-8000-00805**)
UUID: Generic Attribute Profile (1801--1000-8000-00805**)
UUID: Vendor specific   (fe25c237-0ece-443c-b0aa-e02033**)


At this point I started Subsurface (under the same user), selected the
Shearwater/Perdix in the import logs dialog, checked the "download via 
bluetooth"
button, and chose the Perdix BT device that Subsurface presented.

But whenever I tried to start the download, the operation failed after a wait
time of ~5 seconds with a dialog stating:

Insufficient privileges to open the device D4:60:3A:**:**:** Shearwater (Perdix)


"strace" did not reveal any "ENOPERM"-failing operations or such.


Regards,

Lutz Vieweg
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Marco Martin 
On Tuesday 04 April 2017, Dirk Hohndel wrote:
> Thanks for catching this and pointing it out. Something we need to fix
> before pushing this out as a public beta, of course.
> 
> I'd love if you have patches - my next two days will be very busy, not
> sure I'll find much time to look into this.

seen what i did wrong, now on my master it's ok.
I've also changed the look of those boxes to be on top so they don't get under 
the virtual keyboard anymore

-- 
Marco Martin
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Dirk Hohndel 
On Tue, Apr 04, 2017 at 09:17:01PM +0200, Marco Martin wrote:
> On Tuesday 04 April 2017, Dirk Hohndel wrote:
> > 
> > Since the paint function was mostly written by a complete idiot who has no
> > clue what he was doing and just kept adding hack ontop of hack (that would
> > be me)... I wouldn't be surprised if it was completely bogus.
> > 
> > So if you have patches that make things simpler, more power to you!
> 
> well, it was kindof necessary, as before the scaling for high dpi was done by 
> hand by the application itself, now Qt takes (in part) care of it. you can 
> find 
> it in my usual personal clone

I'll play with it tonight - need to run to my next meeting now.

/D
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Dirk Hohndel 
On Tue, Apr 04, 2017 at 09:15:27PM +0200, Joakim Bygdell wrote:
> On 4 Apr 2017 21:06, "Dirk Hohndel"  wrote:
> 
> On Tue, Apr 04, 2017 at 07:50:28PM +0200, Joakim Bygdell wrote:
> > The profile issue is still present in master,
> 
> Which profile issue? There are so many. And is that with my APK or with
> something you built yourself? It turns out that I had forgotten to push
> two commits yesterday...
> 
> 
> Built from master this afternoon.
> Scaling issue where only the top left quarter of the profile where visible.

Yes, that would definitely be broken in master until about 10 minutes ago.
Sorry about that. Now pushed, and a new APK (-198) is on the server in
downloads/daily

> > it happens far more regularly when selecting any of the top two dives in
> > the list
> > than any other dive combined.
> >
> > Dive editing is broken for anything that uses the new HintsTextEdit
> > function.
> 
> Can you elaborate on what is broken?
> 
> 
> Any changes to fields which uses the new hints function is never saved nor
> updated when reloading the dive after edit.
> I have a hunch regarding the reason.

Yeah, that would definitely qualify as "broken". Duh. I played with it to
see if the user experience made sense, but I think I never tried to
actually save changes (since I had it set up to use my "real" dive file
which I didn't want to modify). Silly me.

Thanks for catching this and pointing it out. Something we need to fix
before pushing this out as a public beta, of course.

I'd love if you have patches - my next two days will be very busy, not
sure I'll find much time to look into this.

Thanks

/D
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Marco Martin 
On Tuesday 04 April 2017, Dirk Hohndel wrote:
> 
> Since the paint function was mostly written by a complete idiot who has no
> clue what he was doing and just kept adding hack ontop of hack (that would
> be me)... I wouldn't be surprised if it was completely bogus.
> 
> So if you have patches that make things simpler, more power to you!

well, it was kindof necessary, as before the scaling for high dpi was done by 
hand by the application itself, now Qt takes (in part) care of it. you can find 
it in my usual personal clone

-- 
Marco Martin
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Marco Martin 
On Tuesday 04 April 2017, Dirk Hohndel wrote:
> > 
> > Dive editing is broken for anything that uses the new HintsTextEdit
> > function.
> 
> Can you elaborate on what is broken?

one thing i see on a real phone is that it seems to not play well with the on 
screen keyboard, i think if i put it on top of the text field instead of the 
bottom it would work and look better

-- 
Marco Martin
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Joakim Bygdell 
On 4 Apr 2017 21:06, "Dirk Hohndel"  wrote:

On Tue, Apr 04, 2017 at 07:50:28PM +0200, Joakim Bygdell wrote:
> The profile issue is still present in master,

Which profile issue? There are so many. And is that with my APK or with
something you built yourself? It turns out that I had forgotten to push
two commits yesterday...


Built from master this afternoon.
Scaling issue where only the top left quarter of the profile where visible.


> it happens far more regularly when selecting any of the top two dives in
> the list
> than any other dive combined.
>
> Dive editing is broken for anything that uses the new HintsTextEdit
> function.

Can you elaborate on what is broken?


Any changes to fields which uses the new hints function is never saved nor
updated when reloading the dive after edit.
I have a hunch regarding the reason.


Thanks

/D


// Jocke
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Dirk Hohndel 
On Tue, Apr 04, 2017 at 08:48:05PM +0200, Marco Martin wrote:
> On Tuesday 04 April 2017, Dirk Hohndel wrote:
> > This turned out to be an interesting combination of things coming together.
> > Using lring() to round the device pixel ration instead of truncating it is
> > the main culprit, I have added another hack on the existing hack to
> > position the profile correctly and it seems to work for a reasonably large
> > set of possible DPR values that I've tried.
> 
> in my clone i tried to radically simplify the paint function of the profiles 
> widget (pretty much, getting rid of it) and just trusting qgraphicsview to 
> adapt itself to QT_SCALE_FACTOR
> with that value set to 5 or 6 on the desktop, it seems to still render 
> correctly, would need to be tried on android tough

Since the paint function was mostly written by a complete idiot who has no
clue what he was doing and just kept adding hack ontop of hack (that would
be me)... I wouldn't be surprised if it was completely bogus.

So if you have patches that make things simpler, more power to you!

/D
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Dirk Hohndel 
On Tue, Apr 04, 2017 at 07:50:28PM +0200, Joakim Bygdell wrote:
> The profile issue is still present in master,

Which profile issue? There are so many. And is that with my APK or with
something you built yourself? It turns out that I had forgotten to push
two commits yesterday...

> it happens far more regularly when selecting any of the top two dives in
> the list
> than any other dive combined.
> 
> Dive editing is broken for anything that uses the new HintsTextEdit
> function.

Can you elaborate on what is broken?

Thanks

/D
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Marco Martin 
On Tuesday 04 April 2017, Dirk Hohndel wrote:
> This turned out to be an interesting combination of things coming together.
> Using lring() to round the device pixel ration instead of truncating it is
> the main culprit, I have added another hack on the existing hack to
> position the profile correctly and it seems to work for a reasonably large
> set of possible DPR values that I've tried.

in my clone i tried to radically simplify the paint function of the profiles 
widget (pretty much, getting rid of it) and just trusting qgraphicsview to 
adapt itself to QT_SCALE_FACTOR
with that value set to 5 or 6 on the desktop, it seems to still render 
correctly, would need to be tried on android tough


-- 
Marco Martin
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Subsurface-mobile status

2017-04-04 Thread Joakim Bygdell 
The profile issue is still present in master,
it happens far more regularly when selecting any of the top two dives in
the list
than any other dive combined.

Dive editing is broken for anything that uses the new HintsTextEdit
function.


-- 
Jocke
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Anton Lundin 
On 04 April, 2017 - Willem Ferguson wrote:

> I hear that it is only the Perdix air integration (AI)  that
> has BTLE, not the Perdix without AI. I checked the respective user manuals.
> But I am not sure.

The new-er regular "Perdix" is BTLE to, so we need to get hacking on
that BTLE support.


//Anton - Who plans to do it, any time now...


-- 
Anton Lundin+46702-161604
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Dirk Hohndel 
Sorry, forget what I said. Crossed threads.

⁣--
>From my phone​


 Original Message 
From: Dirk Hohndel 
Sent: Tue Apr 04 04:56:56 PDT 2017
To: Lutz Vieweg 
Cc: Subsurface Mailing List 
Subject: Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

The problem is that it is BTLE which we don't support.
We need to figure out how to find it when scanning, and then the communication 
protocol. There is no serial writer for BTLE like there is rfcomm for BT

⁣--
From my phone​


 Original Message 
From: Lutz Vieweg 
Sent: Tue Apr 04 02:08:35 PDT 2017
To: subsurface@subsurface-divelog.org
Subject: Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

On 04/04/2017 10:27 AM, Willem Ferguson wrote:
> Would it be helpful to insert a section on Bluetooth import from the OSTC 
> Sport into the user
> manual? We have text for some Bluetooth computers e.g. Shearwater Petrel.
>
> Does anyone use the Shearwater Perdix?

A friend of mine just bought one.

I might have an opportunity to try Bluetooth connectivity
to that device, but only this evening for the next 3 weeks.

Is there some specific (mis-)behaviour I should be looking for?

Regards,

Lutz Vieweg

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




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

Re: Subsurface-mobile status

2017-04-04 Thread Marco Martin 
On Tuesday 04 April 2017, Rick Walsh wrote:
> > So this is the one remaining Kirigami 2 issue that I am aware of, and
> > something Marco told me on IRC he'd be looking into. It appears that this
> > should still accept a tap on the button, but on high dpr devices like my
> > Pixel XL it's nearly impossible to just tap - everything is recognized as
> > a tiny drag and then ignored.
> > 
> > I agree it is not intuitive.  I can sometimes tap the button, but mostly
> 
> it doesn't work (recognized as a tiny drag).  For some reason, closing the
> menu by tapping the left arrow slide/button is more reliable than opening
> with the hamburger, for me at least, but still doesn't work all the time.
> Tapping the arrows for submenus always works for me.

on kirigami master should work better now

-- 
Marco Martin
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Re: Fwd: Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Lutz Vieweg 

On 04/04/2017 10:27 AM, Willem Ferguson wrote:

Would it be helpful to insert a section on Bluetooth import from the OSTC Sport 
into the user
manual? We have text for some Bluetooth computers e.g. Shearwater Petrel.

Does anyone use the Shearwater Perdix?


A friend of mine just bought one.

I might have an opportunity to try Bluetooth connectivity
to that device, but only this evening for the next 3 weeks.

Is there some specific (mis-)behaviour I should be looking for?

Regards,

Lutz Vieweg

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

Fwd: Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Willem Ferguson 
-- Forwarded message --
From: "Willem Ferguson" 
Date: 04 Apr 2017 8:23 AM
Subject: Re: OSTC Sport bluetooth problem with Linux
To: "Rick Walsh" 
Cc:

>From phone.
Would it be helpful to insert a section on Bluetooth import from the OSTC
Sport into the user manual? We have text for some Bluetooth computers e.g.
Shearwater Petrel.

Does anyone use the Shearwater Perdix?

Kind regards,
Willem

reflect that native works better.
>
> Thank you for the very good suggestion, and for pointing out that the FAQ
still directs users to use the old method.

Cheers,

Rick


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

Re: OSTC Sport bluetooth problem with Linux

2017-04-04 Thread Anton Lundin 
On 04 April, 2017 - Rick Walsh wrote:

> Hi Stephen,
> 
> On 4 April 2017 at 11:06, Stephen Hemminger 
> wrote:
> 
> > I am running Debian Stretch with current AppImage version of Subsurface.
> > Trying to download dives from OSTC Sport.
> >
> > The Bluetooth pairing is successful and the rfcomm connection comes up,
> > but the serial port won't open.
> >
> > # rfcomm -i hci0 connect rfcomm0 00:80:25:4A:10:77 1
> > Connected /dev/rfcomm0 to 00:80:25:4A:10:77 on channel 1
> > Press CTRL-C for hangup
> > [0.003838] ERROR: Input/output error (5) [in ../../src/serial_posix.c:198
> > (dc_serial_open)]
> > [0.003927] ERROR: Failed to open the serial port. [in
> > ../../src/hw_ostc3.c:344 (hw_ostc3_device_open)]
> >
> >
> > Possibly related dmesg
> > [  482.059744] Bluetooth: TIOCGSERIAL is not supported
> > [  482.060272] Bluetooth: TIOCGSERIAL is not supported
> > [  488.060994] Bluetooth: TIOCGSERIAL is not supported
> >
> > Running stock Debian kernel which is 4.9..
> >
> 
> Instead of rfcomm, have you tried using the "native" Bluetooth support in
> Subsurface to connect to your dive computer.  It is much easier and
> generally more reliable.  See section 5.2.2 of the user manual
> https://subsurface-divelog.org/documentation/subsurface-4-user-manual/
> 
> If that fails, please run subsurface -vv from the command line, try using
> the native Bluetooth support and post the terminal output here.
> 
> Do you have onboard Bluetooth device or a dongle?  Just one device
> connected?  You may need to activate Bluetooth with whatever desktop
> environment tool you have prior to using the Subsurface Bluetooth dialog.
> 
> 
> > PS: Is bluetooth always this much of a a pain or is it just Linux?

Bluetooth comes with its own set of problems.

Another thing to try is a fresh battery in your device. I've seen quite
a few times that a battery which can run the main micro can't power the
bluetooth chip reliably.


//Anton


-- 
Anton Lundin+46702-161604
___
subsurface mailing list
subsurface@subsurface-divelog.org
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface