Routino SVN Repository Browser

Updated for version 2.7.3.

Translation of the final German word missing from the XML file.

Sort the data geographically before pruning so that the data points physically
close together are close together in memory which reduces swapping/paging and
therefore runs much faster when memory is limited.

Comment fixes, tidy-up and one bug fixed.

Use the waysx->idata array instead of wayx->id when logging node errors during
the segment creation.

Correct errors in wayx->id usage when compacting ways.

Improve the message printed at the end when using --logtime or --logmemory.

Clarify some comments.

Use the waysx->idata array when logging duplicate segments rather than looking
up the wayx and using its id, also saves mapping the ways into memory.

Use the nodesx->idata array when logging duplicate segments rather than looking
up the nodex and using its id.

Shrink the size of the nodesx->idata array when removing non-highway nodes.

Free memory that it allocated by IndexSegments() when no longer needed rather
than holding on to it.

Speed up the database generation when pruning by reducing the amount of random
memory accesses are required by compacting the database after each step.  Unless
there is enough RAM to hold all the memory mapped files this should be faster.

Try to speed up the search for the closest node/segment by minimising the number
of nodes that are examined in detail.

Record the memory used by the node, segment, way and relation caches in the slim
mode router.

Log the router time in microseconds rather than milliseconds.
Add a note at the end about the format of the time and memory logging.

Add the '--logtime' and '--logmemory' options to the router to report the time
and maximum memory in use (allocated and mapped files) during each step of the
routing.

Use exit() when exiting the program other than at the end of the main function.

Add a '--logmemory' option to planetsplitter which will report the maximum
memory in use (allocated and mapped files) during each step of the processing.

Be more consistent in the way that cache.h is included.

Make a function static in logging.c rather than global.

Free the segment list before generating the errorlog files (it isn't used).

Change a comment and the style in which some data is freed.

Free segmentsx->usedway in FreeSegmentList() rather than letting it leak.

Free segmentsx->firstnode in SaveNodeList() like DATALIFE.txt says it should be.