summaryrefslogtreecommitdiff
path: root/colm
Commit message (Collapse)AuthorAgeFilesLines
* Function name changes related to pushing and popping ignores.Adrian Thurston2012-05-173-28/+28
|
* Call the detach of the right ignore.Adrian Thurston2012-05-171-1/+10
|
* The prepParseTree function can go away.Adrian Thurston2012-05-172-18/+0
|
* Copy the trees being written to when attaching and detaching ignores.Adrian Thurston2012-05-171-32/+35
|
* Use the parse trees in the bindings array. Probably can turn this into a bit inAdrian Thurston2012-05-176-38/+33
| | | | the parse tree, but that is a change for another day.
* Some minor cleanup.Adrian Thurston2012-05-162-12/+9
|
* Don't need to copy flags from data tree into parse tree.Adrian Thurston2012-05-162-8/+0
|
* Call the parse tree flags PF_*.Adrian Thurston2012-05-165-47/+47
|
* Removed the AF_PARSE_TREE flag, no longer necessary.Adrian Thurston2012-05-164-24/+3
|
* AF_PARSED is no longer used, removed.Adrian Thurston2012-05-163-19/+13
|
* ARTIFICIAL and NAMED flags now used only in parse trees.Adrian Thurston2012-05-164-15/+27
|
* Took the args specific to compile-time out of makeTokenWithData. ThoseAdrian Thurston2012-05-163-10/+10
| | | | operations effected afterwards in ctinput.cc.
* Now translating only the parseTree->id to termDup from nonterminals. The dataAdrian Thurston2012-05-163-43/+12
| | | | | | | | | tree id always remains as the nonterminal. The nested comm output needs to be adjusted because it is now functioning correctly. A nonterm was sent as a comment and was not translated back to the nonterm from the dup when the comment was attached. Now that translation doesn't happen at all and so it is correctly a nonterminal.
* Couple fixes for the token.line operation.Adrian Thurston2012-05-161-1/+5
|
* Added token.line for fetching line information.Adrian Thurston2012-05-164-0/+34
|
* Simplified a debug call.Adrian Thurston2012-05-161-8/+1
|
* AF_TERM_DUP now stays out of the data parse tree node. The NAMED and ARTIFICIALAdrian Thurston2012-05-162-15/+11
| | | | | flags not accessed in the fillNodes. Must complete the movement of those flags too.
* Going to move more flags into the parse tree nodes. Verifying that they areAdrian Thurston2012-05-161-12/+39
| | | | accessible as is there with some assertions.
* The IL_ATTACHED flags belong in the parse tree flags, not the data tree.Adrian Thurston2012-05-152-29/+16
|
* Some cleanup.Adrian Thurston2012-05-151-37/+25
|
* More leaks cleaned up. Test suite runs clean.Adrian Thurston2012-05-142-8/+13
|
* Refcounting improvements.Adrian Thurston2012-05-141-49/+44
|
* Don't need to mirror Tree anymore. Removed some ParseTree fields.Adrian Thurston2012-05-143-24/+4
|
* Parse tree reaping.Adrian Thurston2012-05-141-17/+25
|
* Eliminated PtKid. Worked on first test run (!).Adrian Thurston2012-05-149-222/+191
|
* Eliminated some casts that are no longer necessary.Adrian Thurston2012-05-141-7/+6
|
* Eliminated the pt macro.Adrian Thurston2012-05-149-159/+138
|
* Need the getRhsVal for the C interface.Adrian Thurston2012-05-141-1/+12
|
* Pointers to parse trees in the pda run are now using the parse tree structsAdrian Thurston2012-05-146-75/+46
| | | | (and Kid2).
* Warn about lost trees/kids only if enable-debug has been specified.Adrian Thurston2012-05-141-7/+10
|
* Added a separate Kid data structure for use with parse trees.Adrian Thurston2012-05-145-23/+49
|
* Separate the attach ignores into right and left attach. The right ignores areAdrian Thurston2012-05-131-44/+91
| | | | | | | now attached when a token induces a reduction, as well as on a shift. Don't need to use tokenList, which was unsafe and incorrect when the tree is transformed. The detach was also separated, but we are not yet detaching the rights. All tests pass regardless.
* Only advance reductions where we need to for parse_stop.Adrian Thurston2012-05-134-2/+44
|
* Some code movement.Adrian Thurston2012-05-122-96/+95
|
* Simplified the restore of accumulated parse trees when they are detached. OnlyAdrian Thurston2012-05-122-25/+19
| | | | a single walk.
* Reverse the accum ignore list parse trees when attaching. This is consistentAdrian Thurston2012-05-121-5/+12
| | | | with the reversal of the data trees, as well as reduced parse and data trees.
* Detach parse trees and data trees on reduction. Only the real stack retainsAdrian Thurston2012-05-121-15/+18
| | | | | shadow pointers. Likewise for ignores. When they are attached to the data trees they are detached from the parse trees that maintain the unparsing history.
* Merge branch 'master' of ssh://complang.org/var/git/colmAdrian Thurston2012-05-123-34/+25
|\
| * Use data trees in the token list, not parse trees.Adrian Thurston2012-05-121-12/+12
| |
| * Use AF_HAS_RCODE only in parse tree flags.Adrian Thurston2012-05-113-22/+13
| |
* | Fill nodes using the data tree.Adrian Thurston2012-05-111-60/+58
| |
* | Going to eliminate the use of state in when building patterns and replacements.Adrian Thurston2012-05-111-55/+64
|/ | | | | Want to operate on the data tree. Compute node indexes as we go rather than filling in one pass and referencing in another.
* Better separation between parse trees and data trees in the ignore accumulator.Adrian Thurston2012-05-101-27/+28
|
* Getting stricter about the difference between parse trees and data trees. NeedAdrian Thurston2012-05-103-29/+60
| | | | the restore RHS instruction back. Previously relied on the shadow too much.
* Cleanup the building of child lists in the reduction.Adrian Thurston2012-05-101-27/+15
|
* Warning elimination.Adrian Thurston2012-05-103-16/+29
|
* Fixed some warnings. Removed some old ignore-in-parse-tree code.Adrian Thurston2012-05-103-18/+6
|
* Some cleanup.Adrian Thurston2012-05-091-20/+26
|
* The data trees are now allocated as plain trees, not parse trees.Adrian Thurston2012-05-084-24/+12
|
* Minor cleanup.Adrian Thurston2012-05-081-4/+4
|
View Lorry Controller Status