robthebold | # 1931 | 2017-03-06, 16:19 | Report

Any word on the 'black screen of no maps' on oneplus x? I'm seeing the issue and woukd br happy to help with testing.

Thanks! Other than not seeing maps in latest version I love this program! (I know, 'other than that how was the play mrs lincoln?')

MartinK | # 1932 | 2017-03-14, 22:13 | Report

I sure know about this - at the moment I'm preparing a modRana bugfix release (should fix the map layer switching bug) and after that I plan to return my attention to debugging the 1080p-black-screen issue, primarily in the form of adding more debugging messages, so I can track the issue down. In any case I suggest watching (or even subscribing to) this issue on GitHub:
https://github.com/M4rtinK/modrana/issues/186

That's at the moment probably the best place where to watch for progress in this matter, even though I'll probably also post updates to this thread.

And sorry for this taking so long - the issue seems rather hard to crack so far. My current theory is that it's either some pretty particular race condition (happens at 1080p only, not above or below & only on Sailfish OS) or even possibly some platform adaptation glitch modRana manages to hit on the ex-Android 1080p devices running Sailfish OS.

Actually the two devices seem almost identical in screen (1080x1920), SoC (Snapdragon 801) and GPU (Adreno 330), so I guess a similar hardware adaptation/drivers are used for both ?

It would hardly be the first Sailfish OS specific weirdness I've hit during modRana development:
[bug] MouseEvent x and y properties are floats
[bug] Some QtQuick Canvas drawing coordinates cause application-wide deadlock

BTW, so far the issue seems to happen on Fairphone 2 and OnePlusX - could someone with Sailfish OS running on a phone with 1080p screen verify if current modRana runs fine or if it has similar issues ? That could help narrow the search for a cause down.

If it also happens, it would be more likely some modRana side algorithm is at fault. If modRana works fine then it would be likely specific hardware adaptation glitch is at fault.

mosen | # 1933 | 2017-03-14, 22:18 | Report

Also on Nexus 5 i get black map screen.

MartinK | # 1934 | 2017-03-19, 16:15 | Report

modRana 0.54.4 has been released!

This is mostly a bugfix release fixing various bugs.

Changelog

Black map screen on ported 1080p devices

Next I plan to finally get back to debugging of the map-is-black-on-ported-1080p-devices issue. This release does not have any fixes specific for that, but it would still be helpful if someone who is affected could try to update to 0.54.4 and see if the black-map issue still persists & behaves the same (map is visible for a split second when device is rotated, etc.).

It's quite possible it's a race-condition of some sort, so the various fixes, even though not directly related, could influence it.

Also, is there already someone running Sailfish OS 2.1.0 one a 1080p ported device ? It would be possibly quite interesting to know if the updated Qt 5.6 in 2.1.0 changes behavior of the black-map-screen issue.

Edit: 19.03.2017 18:01
Looks like the issues no longer manifests on ported devices with Sailfish OS 2.1.0/Qt 5.6!

mal, the heroic Sailfish OS porter reports that the black-map issue no longer happens on Fairphone 2 & OnePlus X as long as they are running Sailfish OS 2.1.x, while the issue can still be reliably reproduced with Sailfish OS <2.1.x.

Thanks yet again to mal for helping to find the cause (Sailfish OS <2.1.x & old Qt) & solution (update to Sailfish OS 2.1.x)!

---

Last edited by MartinK; 2017-03-19 at 17:10.

MartinK | # 1935 | 2017-03-25, 12:39 | Report

modRana 0.54.5 has been released!
It is now possible to set opacity for routes, tracklogs and the track logging trace, so that you can see what's underneath while still seeing the route/tracklog/trace.

Also thanks to AL13N for suggesting this in the Jolla Store comments!

Changelog

handaxe | # 1936 | 2017-03-27, 09:22 | Report

Perhaps link a deb here for n900 users and ask for testing?

Or declare the n900 development over, which is understandable and your absolute right. Otherwise, the device sits in limbo....

My 2c worth :-)

HA

MartinK | # 1937 | 2017-03-27, 11:13 | Report

For quite some time I've been using the Maemo Autobuilder to create deb packages and those go directly to Extras Devel, from where I expect most people get their software nowadays. And I'm afraid I'm no longer setup to create correct Maemo-style deb files locally - that's why I want to test modRana at least bit before submitting a new build to autobuilder.

On the other hand, you can just grab the latest tarball:
http://modrana.org/tarballs/modrana/...-0.54.5.tar.gz

Unpack it and then run modrana.py from the resulting directory - it should detect it's running on the N900 and behave accordingly. If you can help test modRana like this it would be really helpful! Don't hesitate to report any weird behavior, crashes, etc. And thanks in advance!

justmemory | # 1938 | 2017-03-27, 11:47 | Report

Hi, MartinK!

First of all, thanks for the update!

Tried this version with my n900 and first major problem is that in this version offline routing is not working however at the same time it is working OK with version 0.53.1 (last official version for Maemo).

jm

EDIT:// running from terminal the relevant section is maybe:
DEBUG mod.route: routing provider ID: Monav
ERROR mod.route: unknown rounting provider ID: Monav

---

Last edited by justmemory; 2017-03-27 at 12:08. Reason: Added log info

justmemory | # 1939 | 2017-03-28, 08:31 | Report

Hi MartinK!

I managed to get offline routing work (tried with foot, car, bicycle) - well almost. Address to address and handmade still does not work.

What I did:
- changed routing provider from Monav to Monav server ("Monav" stuck from earlier version of modrana I guess as it now allows only "Monav server" and "Google online"...)

- /core/bundle/signals_pb2.py - comment out line 6 (from __google.protobuf import descriptor_pb2) as it claimed that cannot import it.

- /core/monav_support.py - changed line 25: import simplejson as json (a "try" - "except" method would be better, it is just a simple solution) as it claimed that cannot import json.

- /core/routing_providers.py - changed line 60: result = self._monav_directions(waypoints, route_params) - as it claimed that it takes 3 argument but provided 2.

- /core/way.py - changed all node/Node.lat and node/Node.lon to node/Node.latitude and node/Node.longitude as it claimed that node/Node has no attribute lat.

- /core/instructions_generator.py - changed all node/Node.lat and node/Node.lon to node/Node.latitude node/Node.longitude as it claimed that node/Node has no attribute lat.

For address to address routing I get the following error: waypoints = [x.getLL() for x in waypoints] AttributeError: 'str' object has no attribute 'getLL'. This is valid as for waypoints it gets the real addresses I add in modrana so it cannot convert it to latitude and longitude points.

For handmade routing I get the following error: Modules/mod_route/mod_route.py, line 577, in _handle_routing_result_cb; if result.route and result.returnCode == constants.ROUTING_SUCCESS:; AttributeError: 'tuple' object has no attribute 'route'.

jm

---

Last edited by justmemory; 2017-03-28 at 08:45.

MartinK | # 1940 | 2017-03-28, 23:25 | Report

Wow, thanks a lot! I've used your pointers and fixed the Monav Server based offline routing on the N900. I've also released an updated tarball, so you - and anyone else who wants to help with testingt - can check it works correctly now:
http://modrana.org/tarballs/modrana/...-0.54.6.tar.gz

As for the other issues:

Address routing - that's actually expected behavior, as Monav does only point to point routing (+ waypoints along the route, IIRC) based on geographic coordinate pairs. It does not support gecoding (turning addresses to coordinates), so it can't work for address-to-address routing when the device is offline. But of course a sensible error message should be shown in such a case, not a traceback.

Handmade routing (basically, "draw your own route" and let modRana follow it) - it indeed seems pretty broken, not just on the N900, but also when running the GTK GUI on desktop. And it looks like it has been this way for quite some time. Will have to look into it more later.