summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Don't flip optimized PNG writes.Tavian Barnes2009-10-301-2/+3
|
* (Sort of) parse boxes.Tavian Barnes2009-10-3010-27/+360
|
* Skeleton of parser and realizer.Tavian Barnes2009-10-296-9/+169
|
* Make sure we die on critical dmnsn_error()'s.Tavian Barnes2009-10-291-0/+2
|
* Correctly tokenize <=, >= and !=.Tavian Barnes2009-10-294-22/+55
|
* Support all POV-Ray keywords.Tavian Barnes2009-10-294-11/+1284
|
* Correct locale "transaction" handling in dmnsn_tokenize().Tavian Barnes2009-10-291-8/+8
|
* Rename DMNSN_T_EXCLAMATION to DMNSN_T_BANG.tavianator2009-10-292-43/+43
|
* Refactor tokenizer a bit.tavianator2009-10-295-28/+93
|
* Don't modify lines, cols, or next in tokenizer helper functions.Tavian Barnes2009-10-291-175/+207
|
* Even more tokenizer work.Tavian Barnes2009-10-293-117/+96
| | | | | | The tokenizer is now able to tokenize all of POV-Ray 3.6's include files. It turns out that some #includes may be dynamic (ex. with #writes beforehand), so the tokenizer can't directly perform includes - the executor must do this.
* Prefix tokens with DMNSN_T_.Tavian Barnes2009-10-292-143/+143
|
* Tokenize all language directives.Tavian Barnes2009-10-292-5/+77
|
* Complete string parsing.Tavian Barnes2009-10-293-9/+71
|
* Set the locale to `C' during tokenization.Tavian Barnes2009-10-281-5/+27
|
* Search in the same path as the current file for #includes.Tavian Barnes2009-10-282-8/+28
|
* Split tokenizer.sh into multiple tests.Tavian Barnes2009-10-287-48/+138
|
* Store source filename in tokens, give better diagnostics.Tavian Barnes2009-10-283-49/+76
|
* Make '#include' work.Tavian Barnes2009-10-282-1/+34
|
* Tokenize strings and language directives.Tavian Barnes2009-10-285-3/+211
|
* Add a vector to numeric tokenizer test.tavianator2009-10-282-1/+4
|
* Tokenize keywords.Tavian Barnes2009-10-285-13/+89
|
* Tokenize identifiers.Tavian Barnes2009-10-282-6/+44
|
* Fix line calculation in comment tokenizer.Tavian Barnes2009-10-281-0/+1
|
* Split some tokenizing work out into helper functions.Tavian Barnes2009-10-281-56/+82
|
* Ignore comments in tokenizer.Tavian Barnes2009-10-274-6/+87
|
* Add line and column numbers to tokens.Tavian Barnes2009-10-272-1/+10
|
* Tokenize numeric values.Tavian Barnes2009-10-275-4/+77
|
* Support more symbols in tokenizer.Tavian Barnes2009-10-276-34/+93
|
* Make tokenizer a bit more robust.Tavian Barnes2009-10-271-3/+19
|
* Use the parallel-tests automake option.Tavian Barnes2009-10-273-4/+8
|
* Fix tokenizer test in VPATH build.Tavian Barnes2009-10-273-3/+4
|
* New tokenizer test.Tavian Barnes2009-10-275-5/+48
|
* New mmap-based tokenizer.Tavian Barnes2009-10-271-13/+33
|
* Begin tokenizer.Tavian Barnes2009-10-264-5/+197
|
* Add some command-line options to `dimension' program.Tavian Barnes2009-10-261-1/+64
|
* Begin work on `dimension' client program.Tavian Barnes2009-10-265-0/+56
|
* Major dmnsn_kD_splay_search() optimization.Tavian Barnes2009-10-261-19/+21
| | | | | | | | At each level of recursion, we have to go down the right branch if it exists. But if we do this before we test the current node and the left branch, we can eliminate those tests in the likely case that we find a closer object in the geometrically larger right subtree. This gives about a 2X speed improvement according to `make bench'.
* Clean up some dmnsn_new_*() functions.Tavian Barnes2009-10-264-64/+37
| | | | | Rather than special-case every failed memory allocation, just make dmnsn_delete_*() more robust and call it.
* Delete empty cube.c.Tavian Barnes2009-10-261-23/+0
|
* Call dmnsn_done_progress() later in dmnsn_finish_progress().Tavian Barnes2009-10-261-3/+2
|
* Don't check if it's too late to add canvas optimizers.Tavian Barnes2009-10-264-26/+6
| | | | This fixes some errors displayed by concurrency checkers like helgrind and drd.
* Fix some concurrency mistakes.Tavian Barnes2009-10-262-12/+31
| | | | TODO: eliminate the too_late field on canvases.
* Speed up dmnsn_kD_splay_deepest_recursive() a bit.Tavian Barnes2009-10-261-8/+7
|
* Fix make distcheck.Tavian Barnes2009-10-252-4/+9
| | | | | Tests and benchmarks were using $(top_srcdir) instead of $(top_builddir) to refer to libraries, and PNG files were not being cleaned.
* Properly generate random bounding boxes in kD splay tree benchmark.Tavian Barnes2009-10-251-13/+34
|
* Benchmark dmnsn_kD_splay().Tavian Barnes2009-10-251-0/+45
|
* Upgrade benchmarks to latest libsandglass.Tavian Barnes2009-10-233-132/+53
|
* Add kD splay tree benchmark.Tavian Barnes2009-10-232-2/+139
|
* Fix NULL assignment to integer value in test library.Tavian Barnes2009-10-221-1/+1
|