Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Changes in double / coordinate printing #570

Closed
wants to merge 69 commits into from
Closed

Changes in double / coordinate printing #570

wants to merge 69 commits into from

Conversation

Algunenano
Copy link
Member

  • Use the shortest representation (enough to guarantee roundtrip).
  • Uses scientific notation for absolute numbers smaller than 1e-8. The previous behaviour was to output 0 for absolute values smaller than 1e-12 and fixed notation for anything bigger than that.
  • Uses scientific notation for absolute numbers greater than 1e+15 (same behaviour).
  • The precision parameter now also affects the scientific notation (before it was fixed [5-8]).
  • All output functions now respect the requested precision (without any limits).
  • The default precision is the same (9 for GeoJSON, 15 for everything else).

Unit tests in https://github.com/postgis/postgis/compare/master...Algunenano:shortest?expand=1#diff-5e03e8529b80411818c3dc29cf2f88ebR269

Many regress test changed mainly because of the fixes to the precision parameter, which is now respected as the amount of digits after the fixed point.

@Algunenano
lwprint: WIP shortest with precision
4a74c5f
@Algunenano
Makefile: Rebuild liblwgeom if ryu code changes
4b86a6d
@Algunenano
Shortest: Add some unit tests
61f9cb6
@Algunenano
Initial implementation of shortest + precision
75957e5
@Algunenano
Skeleton for round trip tests
240c156
@Algunenano
Move everything to shortest
a352165
@Algunenano
WIP
665432c
@Algunenano
Ryu to_chars_fixed: Improve decimal digits rounding
984017a
@Algunenano
to_chars_uint64: Use DIGIT_TABLE
39eed55
@Algunenano
test
0ff1eba
@Algunenano
Simplify checks
683c858
@Algunenano
Revert memset
d532b89
@Algunenano
to_chars_uint64: Use digit table
d0d935a
@Algunenano
Try removing trailing zeros handling
a2aec39
@Algunenano
Revert "Try removing trailing zeros handling"
2ae5ba3 
This reverts commit 33b0131.
@Algunenano
to_chars_fixed: Avoid printing -0
3872d02
@Algunenano
Recover d2d_small_int to speed up small ints
9e29fac
@Algunenano
Recover d2d_small_int to speed up small ints
b754557
@Algunenano
WIP minimal benchmarks
dcf87e9
@Algunenano
Adapt unit test with big numbers
dd9a186
@Algunenano
Scientific notation (huge numbers) now prints all significant digits
5782397
@Algunenano
Fix unit test changes due to respecting the precision requested
110b8ee
@Algunenano
Adapt loader tests to request a specific output precision
a00180a
@Algunenano
Ensure it properly handles rounding into the integer part
4ecc464
@Algunenano
Some bench changes
0e92591
@Algunenano
Adapt geohash regress output
33a572f
@Algunenano
Add some initial round trip tests
715b675
@Algunenano
Adapt in_gml tests
2bda30e 
The changes here are due to ST_Transform not retuning exact integer. With the previous
buggy implementation, we weren't returning all the required decimal digits so the
rounding turned them into ints (but they weren't). After fixing that we show
the correct value
@Algunenano
Adapt in_encodedpolyline tests
5a8ab97 
ST_SnapToGrid is NOT a valid way to truncate precision in the output,
since floating point is, by design, innexact as thus you might be
trying to clip to an unrepresentable value. Instead use the ST_AsText
precision parameter
@Algunenano
Adapt lwgeom_regress tests
49eee5e 
Changes in the scientific notation, as it now prints all
significant digits
@Algunenano
Adapt regress/core/sql-mm-multicurve
c4d7adb
@Algunenano
Adapt regress/core/geos39
3907cdc
@Algunenano
Adapt regress/core/offsetcurve
f9d8813
@Algunenano
Adapt regress/core/tickets
ecb1643
@Algunenano
Ryu: Add some more tests and fix one of them
fc4236b
@Algunenano
Ryu shortest: Fix bug when rounding due to precision parameter
97fb297
@Algunenano
Adapt raster regress rt_pixelvalue
e218b8f
@Algunenano
Adapt topology regress addlinestring
6ea5517
@Algunenano
Adapt in_gml tests for older versions on PROJ
8cf89c2
@Algunenano
Small scientific notation test
d673e11
@Algunenano
Avoid rebuilding liblwgeom all the time because or ryu
72bd33d
@Algunenano
Ryu: Respect precision in scientific notation too
f8461a9
@Algunenano
Ryu: Proper rounding for all digits when applying precision
871ccbb
@Algunenano
Fix ubsan
7afcc94
@Algunenano
Remove TODO
6c4c36b 
Let's keep the old behaviour for now
@Algunenano
Add more tests
a82bf2e
@Algunenano
Better naming
bdb261f
@Algunenano
lwprint: Add tests around the extremes
cc2b353
@Algunenano
Ryu: Remove calls to decimalLength17
f8d4597
@Algunenano
Ryu: Micro optimizations
f1e9a11
@Algunenano
Fix rounding bug
e92a82d
@Algunenano
Ryu: Simplify precision handling
1869b14
@Algunenano
lwprint: Use scientific notation for small numbers and avoid -0
a935aa7
@Algunenano
Ryu: Further simplify and optimize rounding
e0d7d39
@Algunenano
Clean up
f6db751
@Algunenano
Update NEWS
5c90707
@Algunenano
Add some tests and comments
bf6c350
@strk strk closed this in 1343712 Jul 28, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
1 participant
@Algunenano