Tippecanoe formatted every double it wrote through milo::dtoa_milo, a
vendored Grisu2. Grisu2 is fast, but it guarantees neither the shortest
digit string nor the correctly rounded one: it only guarantees that what
it prints parses back to the value it came from. In practice it prints a
digit more than necessary about 0.16% of the time, and picks a neighbor
of the correctly rounded digits about 32% of the time.
This ports Russ Cox's fpfmt (https://github.com/rsc/fpfmt) to C++ in
fpfmt/ and formats through it instead. fpfmt is both shortest and
correctly rounded, and it is faster:
full std::string formatting Grisu2 fpfmt speedup
random bit patterns 156.62 ns 66.83 ns 2.34x
geo coordinates 124.07 ns 58.62 ns 2.12x
short decimals 69.37 ns 49.16 ns 1.41x
small integers 44.18 ns 38.06 ns 1.16x
digit generation only Grisu2 fpfmt speedup
random bit patterns 90.07 ns 20.81 ns 4.33x
geo coordinates 80.64 ns 20.18 ns 4.00x
short decimals 55.61 ns 21.90 ns 2.54x
small integers 40.23 ns 22.50 ns 1.79x
(Intel Xeon @ 2.80GHz, g++ 13.3 -O3. `make fpfmt-bench` reproduces this,
and `./fpfmt-bench -check` reruns the correctness sweep, which is why
milo/dtoa_milo.h is kept even though nothing links it any more.)
The port is deliberately literal, so it can be diffed against fpfmt.go.
Its Short() agrees bit for bit with the Go original's on 445,640 values
covering powers of ten, small integers and reciprocals, subnormals, and
random bit patterns. Over 38.5 million values, fpfmt::dtoa always round
trips, is never longer than Grisu2's output, and is shorter 61,329 times.
Output is otherwise formatted exactly as before, including the choice
between plain and exponential notation, so 26 expected test outputs
change: some numbers lose digits (-26.170044999999999 becomes
-26.170045), and some have a corrected final digit (9.823748927348929e+55
becomes 9.823748927348928e+55). Every changed token was checked to parse
back to the identical double; none of the values themselves moved.
milo/milo.h, whose only job was to declare the C shim jsonpull calls, is
replaced by fpfmt/fpfmt.h, and the shim is renamed dtoa_shortest.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014wJRAuhMninQE4wK2TUfuZ
* Plumb bounding boxes through potential intersections
* Quick bbox reject for bins that can't possibly intersect
* Inching toward attribute accumulation in megatile handling
* Some sort of test for how all these things interact with each other.
Automatic numeric attribute accumulation does *not* apply to attributes
that have an explicit attribute accumulator set, because the order of
operations is too messy and weird
* More sketching
* More sketching
* Actually do some accumulation
* Put all that behind an --accumulate-numeric flag
* Use the same attribute accumulation logic in binning as in megatiles
* Fix backwards conditional
* Add means, but somehow I have some counts of 0
* Handle aggregated attributes with no base attribute in the feature
* Checkpoint before I break everything
* Found a flaw, now to debug
* Fix a typo that broke accumulation
* Add binning tests
* Make sure IDs make it through on the bins
* Fix count/mean accumulation
* Make the numeric accumulation prefix configurable
* Make sure the accumulate test still works with a different prefix
* Forgot to update this test
* More testing to make sure cluster sizes make it all the way through
* Fix neglected --accumulate-attribute when binning
* Mark unexercised attribute accumulation cases as "can't happen"
* Factor out numeric preservation
* Attrs with the accumulation prefix are just preserved, not accumulated
* Test behavior of prefixed attributes
* Plumbing for exclude and exclude-prefix
* Implement and test attribute prefix stripping in overzoom
* Update version and changelog
* For debugging, make an attribute list of source feature IDs
* Revert "For debugging, make an attribute list of source feature IDs"
This reverts commit 65fc99c9d1.
* Start to distinguish fixed cluster density setting from as-needed density
* Make consistent {drop,coalesce}-densest decisions between zooms
* Actually track the previous index instead of just intending to
* Clean up collinearities in coalesced features
* To determine densest, look at actual physical distance, not just index
* Don't actually need the previous index in serial_feature now
* Center of mass of one feature to most distant point of the next
* Add apologetic comment
* Wait, how did the tests pass before?
* Revert "Wait, how did the tests pass before?"
This reverts commit f73c8ee543.
* Add --maximum-string-attribute-length option
* Update version and changelog
* A little more testing to make sure
* Read unidecode data, do some plumbing of it
* More unidecode plumbing
* Do the unidecode smashing, but it doesn't seem to be working
* Ah, that's better!
* Add missing header
* And reorder the includes too
* Shortcut when there is no unidecode data to work with
* Update version and changelog
* Avoid repeated unidecode smashing of the same constant string
* Prep to track conditions other than just "dropped" or "kept"
* Count up instead of down
* Drop or retain whole multiplier clusters based on their first feature
* Calculate a global feature dropping sequence
* Switch over to using the drop sequence for drop-fraction
* Remove unused arguments for the old drop-fraction implementation
* Fix copy-and-paste bugs, update tests
* Properly incorporate feature_minzoom into the drop sequence, I hope
* Rename drop_by to drop_sequence
* See if sorting within clusters fixes filter stability between zooms
* Remove very chatty debug print
* Update changelog and version
* Use named constants instead of numbers for feature dropping/keeping
* Add comment to explain purpose and method of bit reversal
* Add a way to run tippecanoe single-threaded for profiling
* Do less work when the tilestats sample values list is already full
* Save a copy when retrieving the attribute key
* Fewer atomic operations
* Move string hashing from mbtiles to text
* Only do approximate attribute deduplication when writing tiles
* Feature dropping tests are sensitive to exact tile size
* All tile creators now create a string pool for the tile
* Features clipped away to nothing should not participate in that tile
* Revert "Only do approximate attribute deduplication when writing tiles"
This reverts commit c42b34b498.
* Also revert the related test changes
* Revert "Revert "Only do approximate attribute deduplication when writing tiles""
This reverts commit 18509876c3.
* Be more specific about the string hash function
* Use fnv1a instead of std::hash for everything
* Reduce the chance of hash collisions
* Stick a hash search on the front of the tree search in addpool
* Eliminate repeated hashing of the same string
* Switch instead of ifs in json parsing
* A few more cases to populate the hash in addpool
* Store the hash in the tree instead of recalculating
* Add explanatory comment for mysterious argument
* Fewer copies in attribute stringification
* Clean up ancient weirdness in JSON attribute stringification
* More serial_val cleanup
* Pass a serial_feature to rewrite instead of many broken-down arguments
* Get rid of the multiple geometries within `partial`
* Revert "Pass a serial_feature to rewrite instead of many broken-down arguments"
This reverts commit 6f4ab9b725.
* Goodbye, struct coalesce
* Revert "Features clipped away to nothing should not participate in that tile"
This reverts commit 124462fbdc.
* Migrating fields from partial to serial_feature
* Name reconciliation between serial_feature and partial
* Replace struct partial with an augmented serial_feature
* Fix some overzealous search-and-replace renaming
* Don't say struct so often
* Remove more of the former partial construction
* Commenting and cleaning up
* Trying again to avoid all these arguments to rewrite
* I swear I did this same thing before and it didn't work.
* More rewrite cleanup
* Exile --detect-shared-borders to its own file
* Add missing headers
* More commenting and cleanup
* More comments
* Sprinkle consts around
* Emplacing and std::moving
* More cleanup
* That shouldn't have worked after a std::move
* Don't need to allocate memory to compare keys
* Reduce use of the global string pool in tiling
* Another avoidable mvt_value construction
* Further reduction to explicit string pool passing
* These reverses are no longer optimizations
* These layernames can all be references
* Don't drag an unused layername string around with every feature
* Heed a compiler warning about potential buffer overflow
* Fix my confusion about which feature's string pool is relevant
* Avoid some unnecessary allocations in attribute accumulation
* Maybe faster serialization?
* Eliminate a comparison
* Do the same here
* Save a couple of allocations when parsing numbers in JSON
* Immediately assign features to layers instead of subdividing later
* Maintain tilestats for tippecanoe:retain_points_multiplier_sequence
* Crunch out more duplicate attribute values when writing out the tile
* Do tilestats for tippecanoe:retain_points_multiplier_first too
* Shell filters need to be real threads, even if nothing else does
* Simplify tippecanoe_minzoom/maxzoom representation
* Update version and changelog
* Turn off unit, which has bit-rotted
* Instrument line simplification
* Do more work in integer space
* Update tests with simplification done with more integers
* Clean up
* Update changelog
* Eradicate calls to deprecated sprintf()
* Revert "Turn off unit, which has bit-rotted"
This reverts commit 998779313f.
* Fix catch by switching to an unsigned type
* Stop adding features to a tile if it can't possibly work
* Add --integer and --fraction options to tippecanoe-decode
* Carry the strategies field from tileset metadata through tile-join
* Update changelog
* Assign different codes to different kinds of error exits
* Add an option to retain extra coordinate precision at maxzoom
* Make sure not to shift away the extra detail from coordinates
* Add an option to convert double-precision attributes to single
* Sort attribute values in tiles to make them compress a little better
* Slightly improve polygon simplification
By choosing a point that would be retained after simplification
to be the start/end point that always gets retained
* I regret making all of these tests involve polygons
* Add an option to specify the size of tiny polygons
* Fix accidental requiring of argument for --single-precision
* Guard against duplicate points when generating "sizes" for them
* Restore the intended behavior that tiny polygons don't get simplified
* Make the extra detail settable rather than always maximizing it
* Revert "Improve maxzoom guessing for tightly-clustered point data sources (#4)"
This reverts commit fec5e8354c.
* Add an option to prevent choosing a base zoom higher than the maxzoom
* Keep the drop rate high enough when the basezoom gets constrained
* Revert "Revert "Improve maxzoom guessing for tightly-clustered point data sources (#4)""
This reverts commit db6bc27d9e.
* Add --order-by and --order-descending options
* Accept multiple --order-by and --order-descending-by sort keys
* Change JSON objects to a union type to use less memory
* Stop storing the string representation of JSON numbers
* Restore the ability to create features with large integer attributes
* Make sure large-integer feature IDs still behave as before
* Add missing #include
* Don't preallocate as much space for arrays and objects
* Treat inability to check free disk space as a warning, not an error
* Update changelog and version
Claude