Hi Mike, I don't see a simple solution to store the tags in sorted order without a significant increase of the memory footprint. On the other hand, the runtime for sorting is nerly zero compared to the time that is needed to write the data to stderr or the log file.
ciao, Gerd ________________________________________ Von: mkgmap-dev <[email protected]> im Auftrag von Mike Baggaley <[email protected]> Gesendet: Freitag, 10. Februar 2017 11:44:21 An: [email protected] Betreff: Re: [mkgmap-dev] [Patch v1] sorted tags in echotags and log files Hi Gerd, you read my mind – I had been intending to look at that, but not got round to it. Looks good to me. My only question would be whether a similar thing could be achieved and at the same time possibly improve performance by storing the tags in sorted order in the first place (SortedMap?). However, as normally the number of tags is only a handful, I suspect there would be little or no improvement in performance by storing the tags sorted. Cheers, Mike From: Gerd Petermann [mailto:[email protected]] Sent: 10 February 2017 08:47 To: [email protected] Subject: [mkgmap-dev] [Patch v1] sorted tags in echotags and log files Hi all, while comparing results of the patches for the "unpaved problem" I noticed (again) that the echotags function has no specific order, this makes it difficult to compare different outputs of mkgmap. Attached is a patch which changes the way how tags are printed in logs or with the echotags style function. A binary is here: http://files.mkgmap.org.uk/download/333/mkgmap.jar The patch places the tags (key=val) in a SortedSet and uses the toString() method of that set. Effect: - tags are sorted by key - a blank is added as a separator Sample: old: Way 28312451 [highway=path,bicycle=yes,foot=yes,mkgmap:unpaved=1,mtb:scale=1,sac_scale=mountain_hiking,mkgmap:label:1=29,ref=29] new:Way 28312451 [bicycle=yes, foot=yes, highway=path, mkgmap:label:1=29, mkgmap:unpaved=1, mtb:scale=1, ref=29, sac_scale=mountain_hiking] If you analyse the logs with scripts, please check if the additional blanks will cause problems in your tool chain. The patch is quite large because the change allows to remove a lot of code. If I hear no complains, I'll commit this patch next Monday. Gerd _______________________________________________ mkgmap-dev mailing list [email protected] http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
