[Announce] modRana: a flexible GPS navigation system - Page 34 - maemo.org

Wikiwide | # 331 | 2010-11-19, 00:52 | Report

Originally Posted by MartinK View Post

OK, everything ticketed

Well, Woody is currently cooking a on-device tileserver, so I would say that things are moving in the right direction

Glad to hear! Offline routing is welcome.

I like these options:

Originally Posted by MartinK View Post

[*]an arrow pointing from the current position indicator (TangoGPS does it like this)[*]just drawing a long line from current position to a given point[*]or maybe drawing your own route on the map (might be also useful to send to others) ?

What about setting redraw interval, turning automatic redrawing on/off, etc?
You see, it's not necessary to redraw every one second, especially when you are going to blank/lock the screen.
For now, maybe, I'll just keep location-test-gui open for GPS lock, and start Modrana only when I need to see the map.

MartinK | # 332 | 2010-11-19, 01:22 | Report

Originally Posted by eachna View Post

Not the original poster, but, I had the same problem. I had read this thread and much of the wiki, downloaded the maps, and took my n900 with me while driving to jury duty today. I had a vague idea of where to go but didn't know the specific route.

Opened the modRana route option, entered the route, it showed on the map, but I couldn't find a way to "start" the navigation routing. Finally pulled over to the shoulder of the road and stuffed the sim back into my Android phone. Google nav got me to jury duty.

While I had read much of the documentation and played with modRana at home, I hadn't navigated with it until I was in the car. Once I was in my car, I wasn't really in a position to dig through either this thread or the wiki to figure out how to activate the navigation.

When I got home I set up a route to my local deli and walked there. The navigation route worked fine at that point and I can see how to turn it on.

I do realize this is free, alpha software and you're only just starting to polish things up. But, it would be REALLY REALLY helpful to be able to turn on navigation from the Route page (rather than pre-program the Route and start it through "Route => Current Route => Tools".)

If POI is meant to be the way to type in addresses and get to them on the fly, then, it should (somehow) be made clear through the screens. Right now, I don't see anything that looks like it's meant for "on the fly" navigation.

There needs to be a way to suddenly activate GPS routing at a minimum for those "oh damn" moments when you're shifted to a detour you don't expect (edit: and didn't have modRana running till you were detoured -- I don't mean re-routing an existing session). Or, when you're out at a place you do know and are asked unexpectedly to drive somewhere you don't know.

I have added a "autostart navigation" toggle to options->navigation in the last version. It is off by default, but I think I'll change this to on in next release as there is really not much use for a route without directions + the current way of starting navigation is too complicated. But it might still scare a few users, when their N900 suddenly starts talking

Originally Posted by eachna View Post

Related to this, a way to *easily* chose start and end points (either for routes or POI) from your Contacts list addresses would be awesome. Something simpler than typing the address in by hand, which again necessitates pulling the car over when it may not be safe/convenient. I'm borrowing this from Android's Google Nav, which is the service I've been using. It doesn't seem like all that big a deal until you start to use it and it seeps into your GPS-expectation-subconscious

.

Originally Posted by Wikiwide View Post

It's also a feature in Sygic 10, if I'm not mistaken.
Maybe, just add Category "Contacts" to POI and refresh the list each time it is opened?

Hmm, I like this idea

Accessing the contact database might be a bit convoluted but possible. And as Wikiwide suggests, adding this as a special POI category seems logical.

Originally Posted by eachna View Post

Again, I do realize this is development software that one person is working on, not one of Google's pet "eternal beta" products that has a dedicated team updating it on a full-time basis. Reading over this thread I can see how modRana has been progressing over the past few months to really seem like it's catching up to Google Nav in performance.

Awesome software, hope you'll consider the suggestions, off to click the paypal support link!

Thanks! Your support is much appreciated!

The Following 2 Users Say Thank You to MartinK For This Useful Post:
BLC, Wikiwide

白い熊 | # 333 | 2010-11-19, 06:24 | Report

Originally Posted by MartinK View Post

This is a known issue (Czech street names pronounced with English pronunciation sound silly, trust me

), the driving directions that are provided by Google are just html formated strings and the street names are not delimited in any (parsable) way. But it might be possible to do the future, with some other routing service that provides more granular directions, preferably with street names that have language codes

Hmmm... in that case I'd just recommend testing all parts of the text that you can split up for being unibyte/multibyte, I programmed this for sending SMSes in GNU Emacs, since when sending for instance Czech text, or Russian, the length of the SMS is less than half the regular one...

Anyhow, since currently modrana doesn't navigate in Russia due to getting stuck on multibyte text (and I'm sure this is a case in some other countries as well), can you as a temp solution, basically parse the results you get from Google by testing each respectable substring, delimited by spaces for instance, to see if it's multibyte or unibyte? This is easy to do by comparing string length and text length, if they differ, it's multibyte.

If it's multibyte just throw it out and not have espeak pronounce it, and that way it's not gonna get stuck in an endless loop, and there'll be routing navigation in Russia?

Temp solution, but better than not being to route at all currently...

The Following User Says Thank You to 白い熊 For This Useful Post:
Wikiwide

Pigro | # 334 | 2010-11-19, 10:55 | Report

Originally Posted by MartinK View Post

This is indeed strange as both modRana and RR (most probably, did't look to the RR source yet) use Liblocation, which should is provided especially to enable concurrent GPS access.
What was the position indicator in modRana doing ? Stuck in place or rotating or something else ?

Hi Martin. When I started my test drive, I enabled both app's before setting off. modRana 1st, programmed to give me navigation to my destination, then RR turned on and left in background. ModRana spoke initial directions OK, but after RR turned on, the position didn't update as I drove (I hit reroute a few times, but no joy). About 1.5km into drive, I played around a bit at traffic lights, pulling RR to foreground then back to background, and navigating a few modRana menus. Not sure what I did, but after that, modRana had updated my position to near(ish) to where I'd just been, but again stopped updating thereafter.

Once I'd passed the speedcam which was the point of my test (FYI, RR didn't alert me, but that may well be down to the quality/accuracy of the speedcam POI database that I pulled off t'internet to test, still need to check that ..) I closed down modRana.

This is all from memory. Next time I'm in the car I'll repeat to see if it's consistent or (hopefully) just a one-off.

great news that POI proximity is already in your to-do list! I remain flabbergasted at how you find time to be so communicative on this thread amidst all the dev effort. Much appreciated, as always

The Following User Says Thank You to Pigro For This Useful Post:
Wikiwide

MartinK | # 335 | 2010-11-19, 12:43 | Report

Originally Posted by 白い熊 View Post

Hmmm... in that case I'd just recommend testing all parts of the text that you can split up for being unibyte/multibyte, I programmed this for sending SMSes in GNU Emacs, since when sending for instance Czech text, or Russian, the length of the SMS is less than half the regular one...

Anyhow, since currently modrana doesn't navigate in Russia due to getting stuck on multibyte text (and I'm sure this is a case in some other countries as well), can you as a temp solution, basically parse the results you get from Google by testing each respectable substring, delimited by spaces for instance, to see if it's multibyte or unibyte? This is easy to do by comparing string length and text length, if they differ, it's multibyte.

If it's multibyte just throw it out and not have espeak pronounce it, and that way it's not gonna get stuck in an endless loop, and there'll be routing navigation in Russia?

Temp solution, but better than not being to route at all currently...

OK, I think I'll do this:

add an option to remove all non-ASCII words from the directions
split the talking and drawing code (the talking code is currently inside the drawing code, which is not good and might be at least partially causing the problems you describe )
passing the messages to espeak not by command line but as a file in /tmp - might be safer than passing all the Unicode through the shell

BTW, what firmware are you using ? I think I heard that there are some local/global variants.

I am running what appears to be some European version (there are only European languages, including Russian) with the language set to Czech. All the Cyrilic is being displayed correctly (and voice output i working too):

The Following 4 Users Say Thank You to MartinK For This Useful Post:
白い熊, BLC, Wikiwide, xaccrocheur

白い熊 | # 336 | 2010-11-19, 13:27 | Report

Originally Posted by MartinK View Post

BTW, what firmware are you using ? I think I heard that there are some local/global variants.

Wait a second, you have it working... I thought it did get stuck in the cycle with Cyrillic nav everywhere...

Anyhow, mine displays the Russian text fine, just gets stuck in routing.

Hmm... I've upgraded to the latest PR, i.e. 1.3, but since I didn't wanna reflash, just dist apt-get dist-upgrade, anyhow maybe something is wrong as when I check available langs - I only have Czech, English US, English UK, and French. I use US English usually.

Tried switching it to Czech, but the same cycle error when routing.

However, since there are no other available langs, maybe the apt-get upgrade didn't really result in the same state as a proper flash..

Oh no, don't wanna reflash, but maybe I'll have to, as if it's working on your sys, I should be able to get it working. Did you flash to the latest PR 1.3?

Last edited by 白い熊; 2010-11-19 at 13:37.

白い熊 | # 337 | 2010-11-19, 14:00 | Report

Something occurred to me now.

I'm using the power kernel, with the biggest part of the internal drive in an encrypted partition. I mount it after boot and my MyDocs symlinked inside the partition, so all the stuff in MyDocs resides in the encrypted partition...

I've experienced probs in the past with some 'freak' progs not working after modprobing the dm-crypt module which is necessary to mount the partition.

I just tried now without having the partition mounted, started modrana and it DOES route in Russian, though it's freaky as it's a bunch of `in 100 meters turn on letter code 4 a, letter code 5 e, letter code b a...' as it substitutes for the Unicode, I'd assume this'd super uncomfortable having to listen to this while driving, so your idea of splitting it up, is good anyway. However it does route...

I'm gonna experiment some more now, to see if it's the modprobing the module, or having MyDocs symlinked or what's causing the infernal loop...

Any idea why modprobing dm-crypt could have an effect on this loop?

Or it might be overclocking, though it shouldn't, but that's what I use the power kernel for... Anyhow, will try to determine which particular config step it is, after which modrana gets in the routing loop...

MartinK | # 338 | 2010-11-19, 14:16 | Report

Originally Posted by 白い熊 View Post

Oh no, don't wanna reflash, but maybe I'll have to, as if it's working on your sys, I should be able to get it working. Did you flash to the latest PR 1.3?

Just used HAM, then fapman when HAM got stuck and apt-get install -f when fapman itself errored out

But it finished ok in the end and it seems to behave as a regular PR1.3.

Originally Posted by 白い熊 View Post

I just tried now without having the partition mounted, started modrana and it DOES route in Russian, though it's freaky as it's a bunch of `in 100 meters turn on letter code 4 a, letter code 5 e, letter code b a...' as it substitutes for the Unicode, I'd assume this'd super uncomfortable having to listen to this while driving,

Yep, unless the language for navigation is set to Russian (in options->navigation), espeak just reads the Unicode codes of the Cyrilic characters.
Therefore, ASCIIzation is still needed if you want the actual directions in a language other than Russian.
Also, I might try a little heuristics:

check if a sub-string contains Cyrillic
set the language for such sub-string to Russian

Just an idea, might be doable or not

The Following User Says Thank You to MartinK For This Useful Post:
Wikiwide

白い熊 | # 339 | 2010-11-19, 15:02 | Report

Originally Posted by MartinK View Post

Just used HAM, then fapman when HAM got stuck and apt-get install -f when fapman itself errored out

But it finished ok in the end and it seems to behave as a regular PR1.3.

Same what I did...

Originally Posted by

Also, I might try a little heuristics:

check if a sub-string contains Cyrillic
set the language for such sub-string to Russian

Just an idea, might be doable or not

I think this would be quite a good approach actually...

Anyhow, just messing with it, think what's causing it to loop endlessly is the dm-crypt module, need a couple more tries to pinpoint it definitely... However, I don't understand why it should be the case, i.e. if it's interfering with something internally, making some calculation or what have you fail, it should affect all routing, as strange as the idea is, however why can I route in Czech Republic, USA, and not in Russia?

白い熊 | # 340 | 2010-11-19, 15:38 | Report

You're not gonna believe this, I found the problem.

It was /usr/lib/locale

I've installed locales which is not installed by default, but for boot the old pre-locales-install content is needed, so I bindmount the folder after boot.

Turns out I messed up somehow sometime and had the bindmounted /usr/lib/locale empty and for this reason it was cycling forever with the Russian routing, but for some reason working with Czech.

Anyhow, I've restored contents of the /usr/lib/locale dir properly and now it's working.

Ugh... Thanks for help.

Anyhow, please still implement if you can the Cyrillic identification and substitution, as continuous `letter A 6 7 5...' is not too great :O)