[mkgmap-dev] oneway=reverse handling broken.

Fri Jan 22 15:58:01 GMT 2010

On 22.01.2010 16:46, Torsten Leistikow wrote:
> Felix Hartmann schrieb am 22.01.2010 16:24:
>    
>> Well until 1497 auxiliary tags worked, now they are broken. I tried it
>> too. At least if you provide a list of say 15 auxiliary keys. Maybe only
>> one or two are parsed???
>>      
> I can't really comment on the latest changes, since my main styles are depending
> on the correct execution of the action rules, which is only done by the style
> branch.
>
> My styles are much more simple than yours, but nevertheless I will never build
> such complex. Instead of checking
>
> ( highway=primary | highway=secondary | highway=tertiary | highway=minor |
> highway=unclassified )&  ...
>
> in a single rule, I use something like the following:
>
> highway=primary {set road_usable=yes}
> highway=secondary {set road_usable=yes}
> highway=tertiary {set road_usable=yes}
> highway=minor {set road_usable=yes}
> highway=unclassified {set road_usable=yes}
> highway=*&  road_usable!=* {set road_usable=no}
>
> Now in the following rules I just have to check
>
> highway=*&  road_usable=yes&  ...
>
> Based on my experiance this gives much more maintainable style rules. Because if
> I have to change the conditions for the auxiliary tag, e.g. I have to include
> another road type, then I just have to change the rules where the road_usable
> flag is set and all concerned output rules can stay unchanged.
>
>    
Well it was working 100% with 1497. I need to check for all those 
streets because I have 9 different oneway styles, 8 different bridges, 8 
different tunnels, ......
Also many roads have only invisible routing line .....

The biggest problem is that after changing around, I thought it all 
works well and did not do any backups of how my style-file worked before 
until realising the mess. 1498 and upwards if !=* is not working 100% 
correct, is simply unusable. This was a real degradation. Now I will 
need to spend hours trying to get a working style-file for 1497 again if 
soon >1500 is not getting correct behaviour again.

1505 is not doing things in order anymore. If I add a line further back 
in the style, it trashes stuff further to the front. It was never 
possible before to damage lines that were used with [continue] or 
[continue with_actions] with the old way of working. Now you never know 
if you add a line that it does crash your lines in front. Just like the 
conditional rules are working.

I also heavily depend on the conditional rules, and can live pretty well 
with the current behaviour. It just really makes long long lines as you 
have to check for many many combinations.

But now it really means, that a lines file of some thousand lines only 
(1.2MB) becomes unmaintainable because you never know what happens when 
you add some lines. It's like a lottery.
> Gruss
> Torsten
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>    