April 2021 - mkgmap-dev - The mkgmap lists

More logging
by Mike Baggaley 11 Apr '21

11 Apr '21

Hi Gerd, Here is the next patch. This one moves a number of messages from writing to stdout/err to using logging. There are several files changed, but most have only very small changes. I'd be interested to know what the message "Map has tide prediction, please implement" in LBLFileReader means... Cheers, Mike

2 3

Commit r4651: echo.patch by Mike Baggaley
by svn commit 11 Apr '21

11 Apr '21

Version mkgmap-r4651 was committed by gerd on Sun, 11 Apr 2021 echo.patch by Mike Baggaley This brings echo and echotags under logging control. http://www.mkgmap.org.uk/websvn/revision.php?repname=mkgmap&rev=4651

1 0

Re logging improvements
by Mike Baggaley 11 Apr '21

11 Apr '21

Hi Gerd, Please find attached the updated version of the patch. It adds synchronisation of the static variables in StyledConverter and fixes the unit test. It includes the not yet committed changes in codepage.patch, diagnostics.patch and Tests.patch. Cheers, Mike

2 3

Commit r4649: diagnostics.patch by Mike Baggaley
by svn commit 10 Apr '21

10 Apr '21

Version mkgmap-r4649 was committed by gerd on Sat, 10 Apr 2021 diagnostics.patch by Mike Baggaley - diagnostic messages are now written to stderr unless using a logging configuration without a ConsoleHandler - diagnostic messages have the level DIAGNOSTIC when they are logged. Before some had WARNING, some had INFO. - deprecate option --check-routing-island-len, replaced by new options --report-routing-islands and --max-routing-island-len - change documentation for options --report-similar-arcs and --report-dead-ends http://www.mkgmap.org.uk/websvn/revision.php?repname=mkgmap&rev=4649

1 0

diagnostic patch
by Mike Baggaley 10 Apr '21

10 Apr '21

Hi Gerd, Please find attached two patches - one that contains the user interface and logging changes to provide improved output from the diagnostic options, along with the other logging interface changes in readiness for replacing direct writes to stdout/err. The second documents the steps that may be taken to work around code page problems. Cheers, Mike

2 6

Polyline base optimisation
by Ticker Berkin 09 Apr '21

09 Apr '21

Hi Gerd If starting with unsigned deltas in polyline encoding, and attempting to reduce the length by testing reduced x/yBase values, there isn't any point in testing Base-1, as the normal number of bits will be the same (added sign bit) and there will be some overflows. Patch attached to change this - it gives me a very slight reduction in RGN size and might give a performance improvement (I didn't detect any). I've also fixed a couple of log. logic errors and changed minDx/y initial to 0, which is much less confusing because they are symmetric with maxDx/y which do use 0. Ticker

2 6

Commit r4646: (slightly modified) codepage.patch by Mike Baggaley
by svn commit 09 Apr '21

09 Apr '21

Version mkgmap-r4646 was committed by gerd on Fri, 09 Apr 2021 (slightly modified) codepage.patch by Mike Baggaley - improve documentation for logging output http://www.mkgmap.org.uk/websvn/revision.php?repname=mkgmap&rev=4646

1 0

Commit r4645: LinePrepOpt.patch by Ticker Berkin
by svn commit 09 Apr '21

09 Apr '21

Version mkgmap-r4645 was committed by gerd on Fri, 09 Apr 2021 LinePrepOpt.patch by Ticker Berkin - improve efficency of bitstream optimization in LinePreparer "If starting with unsigned deltas in polyline encoding, and attempting to reduce the length by testing reduced x/yBase values, there isn't any point in testing Base-1, as the normal number of bits will be the same (added sign bit) and there will be some overflows. Patch attached to change this - it gives me a very slight reduction in RGN size and might give a performance improvement (I didn't detect any). I've also fixed a couple of log. logic errors and changed minDx/y initial to 0, which is much less confusing because they are symmetric with maxDx/y which do use 0." http://www.mkgmap.org.uk/websvn/revision.php?repname=mkgmap&rev=4645

1 0

Unit Tests Patch
by Mike Baggaley 08 Apr '21

08 Apr '21

Hi Gerd, Please find attached a patch that fixes the unit tests when using logging instead of writing to stderr/out (a few of the tests will fail if used against the current mkgmap release as I have changed some checkNoError() calls to checkError("Number of ExitExceptions: 0") - this was written to stdout, but is written to stderr under logging control. Cheers, Mike

1 0

logging improvements
by Mike Baggaley 07 Apr '21

07 Apr '21

Dear all, Please find attached a first stab at improving the way error and other messages are written, which at present is an ad hoc mixture of writing to stderr, stdout and using Java logging, with no apparent logical structure. Writing to stderr or stdout is not thread safe and causes problems when max_jobs is greater than 1. When stderr is redirected to a file and threads simultaneously attempt to write to it, you end up with extra numbered log files containing a few lines that you would expect to be in the named log file. As mkgmap supports a configuration file that defines how messages should be logged, it doesn't make much sense writing a significant number of messages to stderr which is not under logging control. There are several diagnostic options that write their output using warning as the severity, but as this is not enabled by default in the logging, they appear to do nothing unless a logging configuration file is also used, which is not intuitive. The --check-routing-island-len option seems somewhat confused as it attempts to combine both reporting and fixing depending on the value supplied. It also requires info level of logging which would include a lot of other not necessarily wanted messages. If you are using it to fix, you don't necessarily also want any messages. In my view the way messages should be handled is as follows: If the message is an informational message about the program itself and is unrelated to processing any data then write it to stdout - this would apply to the output of --version and --help. If the message is an error message about an invalid parameter in the command line, write it to stderr. Once the command line has been processed sufficiently that a file is to be processed then write messages using logging. If a diagnostic option is enabled, the program should automatically log the requested diagnostic messages. This would apply to --check-roundabouts, --check-roundabout-flares, --report-similar-arcs, --check-routing-island-len and --report-dead-ends. The attached patch enhances the existing logging interface by providing the following: A mechanism for writing messages without any formatting getting added and that always get written whatever level of logging is enabled. A mechanism for writing stack trace information to the log file (I don't believe stack traces really have any place in released code but have not attempted to replace any of these). Mostly directs messages as per my above proposal, though there is the occasional place where the best option is not quite obvious to me, as I think some of the code is not part of the main mkgmap program (there are several main functions and some of the code is for testing purposes). Only logs the start and finish times if files are being processed. Outputs requested diagnostic messages regardless of enabled logging levels Changes --report-dead-ends to be off by default, with the value of 1 being the default if --report-dead-ends is specified without a value. Filters duplicated --report-similar-arcs messages and ignores synthesised arcs so that it can be used together with --make-opposite-cycleways. Deprecates --check-routing-island-len, replacing it with --report-routing-islands to handle reporting and --max-routing-island-len to handle fixing. I expect there will be the odd place where others may disagree with the logging level for a message, but no doubt we can come to a consensus. The patch touches quite a few source files, but there are only a few changes that are anything more than altering the message outputting mechanism. There is no change to any map outputs. Cheers, Mike

2 14