summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Plug leak in list benchmark.Tavian Barnes2010-05-221-0/+1
|
* Add std-options check to 'make installcheck'.Tavian Barnes2010-05-212-2/+16
|
* Use quiet build and colored tests.Tavian Barnes2010-05-212-15/+17
|
* Use "dimension.h" rather than <dimension.h> in #includes.Tavian Barnes2010-05-213-3/+3
|
* Use mixed declarations in array benchmark.Tavian Barnes2010-05-211-6/+5
|
* Add list benchmark.Tavian Barnes2010-05-212-4/+140
|
* Fix memory handling in dmnsn_new_thread().Tavian Barnes2010-05-211-1/+2
|
* Store the bounding boxes of child PR-tree nodes in the parent.Tavian Barnes2010-05-202-60/+67
| | | | This improves cache locality and is good for about a 10% performance boost.
* New dmnsn_vector_element() function.Tavian Barnes2010-05-181-0/+25
|
* .gitignore libdimension.pc.Tavian Barnes2010-05-171-0/+3
|
* Use /* style comments instead of // in realize.c.Tavian Barnes2010-05-171-2/+2
|
* Add a pkg-config file to libdimension.Tavian Barnes2010-05-133-0/+15
|
* Don't attempt to use profile data for profiling build.Tavian Barnes2010-05-131-1/+1
|
* Fix some copyright dates.Tavian Barnes2010-05-1080-81/+81
|
* Write some more docs, and fix a couple things noticed while writing them.Tavian Barnes2010-05-105-15/+166
|
* New dmnsn_new_thread() function.Tavian Barnes2010-05-096-174/+228
| | | | | | | Two advantages: first, waiters are guaranteed to wake if a dmnsn_error() cancels a thread. Second, the thread library is fully abstracted over by threads.{c,h} and progress.{c,h} now, though raytrace.c is still using pthread_create() directly.
* Clean up some unused variables.Tavian Barnes2010-05-092-4/+1
|
* New DMNSN_ARRAY_FOREACH() macro, faster than iterating with dmnsn_array_get().Tavian Barnes2010-05-0812-218/+169
|
* Handle degenerate cases in ray/box intersections.Tavian Barnes2010-05-082-1/+19
|
* Fix argument to malloc().Tavian Barnes2010-05-071-1/+1
|
* Don't store unbounded objects (planes, etc.) in the PR-tree.Tavian Barnes2010-05-072-24/+94
| | | | | Keep them in a dmnsn_array* instead. This makes the tree better and saves us some search time.
* New dmnsn_array_from_list() function.Tavian Barnes2010-05-072-7/+24
|
* Make tests pass with -fno-signed-zeros.Tavian Barnes2010-05-061-2/+4
|
* Only render in one combined `render' test.Tavian Barnes2010-05-064-83/+178
| | | | GL and PNG tests now only test import/export.
* Get rid of "../" in #includes.Tavian Barnes2010-05-068-8/+8
|
* Use C99 for loop initializers.Tavian Barnes2010-05-0524-177/+127
|
* Write a PR-tree test.Tavian Barnes2010-05-052-2/+63
|
* Optimize ray-AABB intersection tests a bit.Tavian Barnes2010-05-051-51/+29
|
* Make dmnsn_new_prtree() handle empty scenes.Tavian Barnes2010-05-051-0/+6
|
* Fix prtree benchmark.Tavian Barnes2010-05-051-95/+24
|
* Replace BVSTs with priority R-trees.Tavian Barnes2010-05-0510-517/+636
|
* Optimize cube intersection function a bit.Tavian Barnes2010-05-041-59/+82
|
* Add dmnsn_{min,max}() inlines.Tavian Barnes2010-05-041-0/+14
|
* Don't apply the transformations which create an object to its textures.Tavian Barnes2010-05-031-0/+13
| | | | For example, sphere { 0, 2 } would scale the texture by a factor of 2.
* Don't apply object modifiers to a NULL CSG object.Tavian Barnes2010-05-031-1/+2
|
* Don't report success when autogen.sh fails.Tavian Barnes2010-05-031-4/+4
|
* Remove debugging code from dmnsn_list_sort().Tavian Barnes2010-05-031-8/+2
|
* Fix list sorting.Tavian Barnes2010-05-012-9/+25
|
* Implement list sorting.Tavian Barnes2010-04-302-18/+109
|
* Use dmnsn_assert() for array and list range checks.Tavian Barnes2010-04-304-38/+19
|
* Add a generic list type.Tavian Barnes2010-04-306-2/+246
|
* Typo in comment.Tavian Barnes2010-04-271-1/+1
|
* Fix some includes.Tavian Barnes2010-04-2723-3/+24
|
* Fix some spurious includes.Tavian Barnes2010-04-272-4/+3
|
* Give backtraces for diagnostics from macros and include files.Tavian Barnes2010-04-189-27/+66
| | | | | Partly this means that dmnsn_diagnostic() can't be called anywhere except during parsing, because the dmnsn_parse_location::parent pointers dangle.
* Plug memory leak in object parsing failure path.Tavian Barnes2010-04-181-42/+45
|
* Have dmnsn_diagnostic() take a dmnsn_parse_location.Tavian Barnes2010-04-1812-156/+103
|
* Support objects in object{} blocks.Tavian Barnes2010-04-181-0/+22
|
* Support all object identifier types.Tavian Barnes2010-04-181-1/+6
|
* Fix segfault on parse error within macro.Tavian Barnes2010-04-181-1/+2
| | | | | The macro token buffer was being freed by both yylex_cleanup() and dmnsn_delete_patricia_trie().