Mapsforge 0.5

[fzk]

FZK sorry, but I have to say that 0.5 version is really not planned. You still do not answered to my question about compare of speed rendering mainly at high zoom levels. Anyway I believe that adding support for some features included in 0.5 is the way I wants to go, even a slower and more painful.

Sorry for not answering. Yes you are right, the Locus rendering speed is (currenty) better. But I think you overestimate this. And the mapsforge devs know that they have improvement potential concerning this point.

Klaus

PS: I regret to hear that you plan to go a proprietary way.

[Menion]

I understand your point of view, on second side I believe my point of view is also logical.

MapsForge is done as universal library. Because of this, at least 50% of code is useless for Locus. Also many optimizations cannot be done because of this "universality". There is also another aspect - it is still not sure, if this project will be success or not. Looks just on current state - https://github.com/mapsforge/mapsforge - last update 3 months old? What may I expect from project, where is not a change for a three months?

[jusc]

@menion,

the last update is from 31.May 2015 as far as I can see. Please look at the middle of the page.

[Menion]

ah you are right .. tiny changes by author of Cruiser: https://github.com/mapsforge/mapsforge/commits/master

[gynta]

fzk> same zoom level..
gynta> Are you sure?..
fzk> Yes, it's in both cases the same zoom level.. ...Locus hasn't. Both effects are resulting in different looking maps...

Ah thx, i see...
...but i miss the big "AHA-effect"
I don't see huge differences.

Have a look on this images -> Link to compare.
(hope it works on all browser)
Sometimes there are missing house numbers in Locus - but in Cruiser also!
Look eg at this detail on the site

[fzk]

I understand your point of view, on second side I believe my point of view is also logical.

No, it is not logical for me. You don't have to take anything away, you only add value. I will close this discussion from my side. Everything is said, you already have made your decision some month ago. It is what it is.

Klaus

[Menion]

you really think it is so simple to take new MapsForge library and with a few lines of code, attach it to Locus application? Well, sorry. It isn't. I made this decision few years ago, when I started modify old 0.3 library to Locus needs.

You may help me make current system better in similar way like @Tobias. He told me what he wants most - support for styleMenu system. It took me few months, but it is there and it works. In next months, I wants to look on better system of placing labels. If there is something you miss most, simply write me. I'll discuss this with Peter and if it make sense, I'll add that feature into Locus as well. I do not understand where is a big problem here...

Look, I also do not want to complicate life to you as a creator of themes. It do not make sense, to create some extra versions, just because some Menion decided ... agree

[jusc]

Look, I also do not want to complicate life to you as a creator of themes. It do not make sense, to create some extra versions, just because some Menion decided ... agree

I think that is not possible.
I started with openandromaps and Locus and I tried to make it compatible to the Locus store maps.
Most of tags are working in the same way, some do not work correctly, some are not available in one or both maps.
FZK created his own map with special tags that are not available in OSM.
So, what ever you (Menion) implement into Locus maps library , theme developers have to work on theme specialized for this one map.
I think Mapsforge is the lowest unique level of all themes, but not more.

[Tobias]

I think Klaus' idea of having two different mapsforge libraries shows some of the ambivalent way the development has taken the last years with mapsforge.
There are community based maps and themes that are made for as many apps as possible, for those are compatibility and an open standard very important (e.g. OpenAndroMaps, Freizeitkarte).
On the other hand Locus built its own proprietary ecosystem with maps, themes and POI database upon mapsforge. For this compatibility is not a big issue, adapting to the needs of this ecosystem is much more important.
Of course there are more sides to it, but with these agendas in mind one library can't really serve both purposes (I'm a bit reminded of internet browsers and html standards ;-)).

What also developed is the usage of mapsforge maps outside of android apps. People are so used to the way they look and appreciate that, so based on mapsforge now there are tileservers, mobac generates raster maps and there are java apps for PCs like Atlas and Route Converter. So route planing even with QuoVadis is possible now.

So of course I would welcome Klaus' idea, as it would take all this into account, but I also appreciate Menion's efforts to at least have some of the new options.

@menion: Just some clarifications on mapsforge development: yes, it was dormant for a while, but last year two major releases were released. Master, which your referred to above, is the stable branch, development happens here and is quite active:
https://github.com/mapsforge/mapsforge/tree/dev

@jusc: As well as Locus has been optimized, mapsforge has been optimized, too, not only new features, and renders differently. Even pure mapsforge 0.3 renders different than Locus nowadays, as Locus is prioritizing captions to symbols (at least it looks so).
Lots of rendering issues have been fixed in mapsforge 0.5. There are so many differences now, not only device dependent scaling and positions of lines vs. fillings. Other things include dy for symbols/captions on areas, dy on ways/pathText/lineSymbol that are scaled the same as stroke-width, automatic position of captions in relation to symbols, priorities that work very well and are very nice combined with a system to switch on/off collision detection per rule etc.
All this gives me more control to get the map rendered without the pitholes and workarounds that were necessary with mapsforge 0.3, and there is more in development (label layer, which solves another big issue with missing POIs/captions).

@fzk: I made a script to convert my mapsforge 0.5 themes, so they work now well with Locus, if you're interested send me your Email.

[Menion]

Nice summary @Tobias , thanks!

My two cents - everything clear and fully understand. Thanks for link on "dev" branch. I was trying now almost four hours to re-use some major parts of current library (mainly stuff around labels), but it is now almost impossible. Since two years, there is now too many differences between modified library in Locus.

So for now, I see few options:

- completely switch to new library (which cause lost of all improvements, massive reduce of speed and increase of battery consumption, lost of auto-loading and probably some more problems, advantage will be better support for styles)

- invest some more time and modify existing Locus version to better fit current MapsForge lib (probably worst variant as more problems will come later because of same reason as now)

- probably best solution from my point of view, is to wait few more months to see, what this lib and it's developers will do. Ten best should be to invest some of my own improvements and then switch completely to next 0.6 version.

So if you excuse me, for now I really do not wants to change library in Locus, because extra work and lost of speed and some features, are for me now worst then benefits. So we will see later, thanks for understanding.

[Tobias]

- i have locus installed to internal SD and my Vectormaps are stored on extern SD, Elevate theme was installed automatic to internal theme folder. So at first moment Elevate was not available until i´v moved it to external theme folder. Perhaps you can add this info to your instruction file for new users

I just checked, this is fixed with latest updates.
Also the theme switcher works now reliable and remembers the settings. Thanks menion!

[Menion]

At least something, you are welcome!

[michaelbechtold]

Hi Menion,
not only me who suffers from missing priority support in Locus (see link below). And I am not re-opening the discussion about switching the library! But rather urge for selectively adding major useful features that MapsForge 0.5 delivers. And that a number of your competitors already support. Having villages supersede their neighborhood town or cities is simply a mess in Locus.

Same thing with the new MultiLanguage feature in MF 0.5.

http://www.openandromaps.org/en/oam-forums/topic/place-names-priority#post-8981

[Menion]

Good day Michael, understand. So work on this is one of priorities, but not the biggest one. Not sure if I'll find time on it during April, but hope that during May I'll have finally time to work on it. Thanks for understanding.

[kech61]

Hi Menion,

I just tested the Multilanguage feature on Mapsforge with Atlas - this is really great.
Especially for maps covering Russia + Asian/Arab Countrys.
The map I tested was the one of Israel / Syria. Switching between english, arabic, hebrew is perfect fine.

The disadvantage is that these Maps have a different file-format than the V3 maps.
If you could find a way to implement this feature this would add lots of value.