1 Ragel 5.20 - Apr 7, 2007
2 ========================
3 -The cs variable is now always initialized, unless the "nocs" option is given
4 to the write init command. If there is no main machine, cs is initialized to
5 the entry point defined by the last machine instantiation.
6 -A number of fixes were made to the Ruby code generator.
7 -The frontend now scans ruby comments and regular expressions.
8 -A transformation for Ruby was added to the language-independent test suite.
9 The Ruby code generator passes on all the language-independent tests.
10 -A new Ruby test and two language-independent tests were added.
11 -Some portability fixes were made (Patches from Josef Goettgens and Aaron
13 -Fixed a make dependency bug which caused a problem for parallel building
14 (Patch from Jeremy Hinegardner).
16 Ragel 5.19 - Mar 14, 2007
17 =========================
18 -Added an import statement to ragel. This statement takes a literal string as
19 an argument, interprets it as a file name, then scrapes the file for
20 sequences of tokens that match the following forms. Tokens inside ragel
21 sections are ignored. An example is in test/import1.rl
25 "define" name lit_string
26 -Added an export mechanism which writes defines for single character machines
27 that have been tagged with the export keyword in their definition. Defines
28 are used for C, ints for D, Java and Ruby. Examples of the export feature
29 are in test/export*.rl.
30 -All machine instantiations are now always generated, even if they are not
31 referenced. In the backend, entry points for all instantiations are written
32 out alongside start, error and first final states.
33 -If the main machine is not present then do not emit an error. Generate the
34 machine without a start state and do not initialize cs in the write init
36 -Added an option -l to rlgen-cd which inhibits the writing of #line
38 -Added a new syntax for verbose embeddings. This adds parentheses:
40 Verbose embeddings without parentheses can make code difficult to read
41 because they force a space in the middle of an action embedding. There is a
42 tendency to associtate spaces with concatenation. Without syntax
43 highlighting to make it clear that the embedding type is a keyword, the
44 problem is especially bad. The danger is that a verbose embedding could be
45 read as an embedding of the keyword representing the empbedding type. With
46 parentheses, verbose embeddings read much more clearly.
47 -Conditions now have a forced order when more than one is executed on a
48 single character. Previously ordering relied on pointers, which caused
49 results to vary by compiler. Ordering is now done using conditon action
50 declaration order. This fixes the failure of cond4.rl which occured with
51 g++ 4.1 and other compiler versions.
52 -In the port from flex to ragel, the name separator :: in Ragel code was
54 -In the examples directory switched from rlcodegen to rlgen-cd. Silenced a
55 warning in statechart.rl.
56 -In the root makefile the distclean target was fixed. It was calling clean in
57 the subdirs. In docs, the clean target was not deleting the new manpages for
58 the rlgen-* programs. Fixed.
59 -Portability and other fixes from Josef Goettgens were applied.
60 -The @datadir@ and @mandir@ variables are made use of in doc/Makefile.in for
61 specifying where documentation should be installed. Patch from Marcus
64 Ragel 5.18 - Feb 13, 2007
65 =========================
66 -There is now a 100% correspondence between state id numbers in the
67 intermediate XML file, Graphviz dot files and generated code. This was
68 achieved by moving code which determines if the error state is necessary
69 into the frontend, and then assigning state numbers before writing out the
71 -Backened class structure was reorganized to make it easier to add new code
72 generators without having to also modify the existing code generators.
73 -The C and D code generation executable was changed to rlgen-cd.
74 -The Java code generation was split out into it's own exectuable (rlgen-java)
75 to allow it to freely diverge from the C/D-based code generation.
76 -The graphviz dot file generation was also split out to it's own executable
78 -The Ruby code generation patch from Victor Hugo Borja was added. This is
79 highly experimental code and is not yet completely functional. It is in the
80 executable rlgen-ruby.
81 -The problem with large state machine machines in Java was fixed. This
82 problem was discovered by Colin Fleming, who also contributed a patch.
83 Rather than specify arrays as comma-separated lists of literals, array
84 initialization is now done in a static function. This is the approach used
85 by the Java compiler. Unlike the compiler Ragel is careful split large
86 initilization functions.
87 -The manual was expanded and reorganized somewhat.
88 -Eliminated per-example directories in examples/.
89 -Made some fixes to the pullscan.rl example.
90 -In the frontend CR characters are now treated as whitespace.
91 -Updated to the latest aapl. This completely eliminates the shallowCopy
92 function. With that, a definitive memory leak is fixed.
93 -Control codes with escape sequences are now printable characters (-p
94 option). Also, the space character is now printed as SP.
95 -Fixed the null dereference and consequential segfault which occurred when
96 trying to create empty machines with [] and // and /a[]b/.
97 -Fixed the segfault which occured when a machine reference failed.
98 -Discontinuing ragel.spec. It is more appropriate for this to be written by
99 package maintenance developers.
101 Ragel 5.17 - Jan 28, 2007
102 =========================
103 -The scanners and parsers in both the frontend and backend programs were
104 completely rewritten using Ragel and Kelbt.
105 -The '%when condition' syntax was functioning like '$when condition'. This
107 -In the Vim syntax file fixes to the matching of embedding operators were
108 made. Also, improvements to the sync patterns were made.
109 -Added pullscan.rl to the examples directory. It is an example of doing
110 pull-based scanning. Also, xmlscan.rl in rlcodegen is a pull scanner.
111 -The introduction chapter of the manual was improved. The manually-drawn
112 figures for the examples were replaced with graphviz-drawn figures.
114 Ragel 5.16 - Nov 20, 2006
115 =========================
116 -Policy change: the fhold and fexec directives did not function correctly in
117 scanner pattern actions. In this context manipulations of p may be lost or
118 made invalid. In the previous version of Ragel they were banned because of
119 this. Instead of banning these directives they have been fixed. The fexec
120 and fhold directives now manipulate tokend, which is now always used to
121 update p when the action terminates.
123 Ragel 5.15 - Oct 31, 2006
124 =========================
125 -A language independent test harness was introduced. Test cases can be
126 written using a custom mini-language in the embedded actions. This
127 mini-language is then translated to C, D and Java when generating the
128 language-specific test cases.
129 -Several existing tests have been ported to the language-independent format
130 and a number of new language-independent test cases have been added.
131 -The state-based embedding operators which access states that are not the
132 start state and are not final (the 'middle' states) have changed. They
134 <@/ eof action into middle states
135 <@! error action into middle states
136 <@^ local error action into middle states
137 <@~ to-state action into middle states
138 <@* from-state action into middle states
140 <>/ eof action into middle states
141 <>! error action into middle states
142 <>^ local error action into middle states
143 <>~ to-state action into middle states
144 <>* from-state action into middle states
145 -The verbose form of embeddings using the <- operator have been removed.
146 This syntax was difficult to remember.
147 -A new verbose form of state-based embedding operators have been added.
148 These are like the symbol versions, except they replace the symbols:
150 with literal keywords:
152 -The following words have been promoted to keywords:
153 when eof err lerr to from
154 -The write statment now gets its own lexical scope in the scanner to ensure
155 that commands are passed through as is (not affected by keywords).
156 -Bug fix: in the code generation of fret in scanner actions the adjustment to
157 p that is needed in some cases (dependent on content of patterns) was not
159 -The fhold directive, which decrements p, cannot be permitted in the pattern
160 action of a scanner item because it will not behave consistently. At the end
161 of a pattern action p could be decremented, set to a new value or left
162 alone. This depends on the contents of the scanner's patterns. The user
163 cannot be expected to predict what will happen to p.
164 -Conditions in D code require a cast to the widec type when computing widec.
165 -Like Java, D code also needs if (true) branches for control flow in actions
166 in order to fool the unreachable code detector. This is now abstracted in
167 all code generators using the CTRL_FLOW() function.
168 -The NULL_ITEM value in java code should be -1. This is needed for
169 maintaining tokstart.
171 Ragel 5.14 - Oct 1, 2006
172 ========================
173 -Fixed the check for use of fcall in actions embedded within longest match
174 items. It was emitting an error if an item's longest-match action had an
175 fcall, which is allowed. This bug was introduced while fixing a segfault in
177 -A new minimization option was added: MinimizeMostOps (-l). This option
178 minimizes at every operation except on chains of expressions and chains of
179 terms (eg, union and concat). On these chains it minimizes only at the last
180 operation. This makes test cases with many states compile faster, without
181 killing the performance on grammars like strings2.rl.
182 -The -l minimiziation option was made the default.
183 -Fixes to Java code: Use of the fc value did not work, now fixed. Static data
184 is now declared with the final keyword. Patch from Colin Fleming. Conditions
185 now work when generating Java code.
186 -The option -p was added to rlcodegen which causes printable characters to be
187 printed in GraphViz output. Patch from Colin Fleming.
188 -The "element" keyword no longer exists, removed from vim syntax file.
189 Updated keyword highlighting.
190 -The host language selection is now made in the frontend.
191 -Native host language types are now used when specifying the alphtype.
192 Previously all languages used the set defined by C, and these were mapped to
193 the appropriate type in the backend.
195 Ragel 5.13 - Sep 7, 2006
196 ========================
197 -Fixed a careless error which broke Java code generation.
199 Ragel 5.12 - Sep 7, 2006
200 ========================
201 -The -o flag did not work in combination with -V. This was fixed.
202 -The split code generation format uses only the required number of digits
203 when writing out the number in the file name of each part.
204 -The -T0, -F0 and -G0 codegens should write out the action list iteration
205 variables only when there are regular, to state or from state actions. The
206 code gens should not use anyActions().
207 -If two states have the same EOF actions, they are written out in the finish
209 -The split and in-place goto formats would sometimes generate _out when it is
210 not needed. This was fixed.
211 -Improved the basic partitioning in the split code gen. The last partition
212 would sometimes be empty. This was fixed.
213 -Use of 'fcall *' was not causing top to be initialized. Fixed.
214 -Implemented a Java backend, specified with -J. Only the table-based format
216 -Implemented range compression in the frontend. This has no effect on the
217 generated code, however it reduces the work of the backend and any programs
218 that read the intermediate format.
220 Ragel 5.11 - Aug 10, 2006
221 =========================
222 -Added a variable to the configure.in script which allows the building of
223 the parsers to be turned off (BUILD_PARSERS). Parser building is off by
224 default for released versions.
225 -Removed configure tests for bison defines header file. Use --defines=file
227 -Configure script doesn't test for bison, flex and gperf when building of the
228 parsers is turned off.
229 -Removed check for YYLTYPE structure from configure script. Since shipped
230 code will not build parsers by default, we don't need to be as accomodating
231 of other versions of bison.
232 -Added a missing include that showed up with g++ 2.95.3.
233 -Failed configure test for Objective-C compiler is now silent.
235 Ragel 5.10 - Jul 31, 2006
236 =========================
237 -Moved the check for error state higher in the table-based processing loop.
238 -Replaced naive implementations of condition searching with proper ones. In
239 the table-based formats the searching is also table-based. In the directly
240 executed formats the searching is also directly executable.
241 -The minimization process was made aware of conditions.
242 -A problem with the condition implementation was fixed. Previously we were
243 taking pointers to transitions and then using them after a call to
244 outTransCopy, which was a bad idea because they may be changed by the call.
245 -Added test mailbox3.rl which is based on mailbox2.rl but includes conditions
246 for restricting header and message body lengths.
247 -Eliminated the initial one-character backup of p just before resuming
249 -Added the -s option to the frontend for printing statistics. This currently
250 includes just the number of states.
251 -Sped up the generation of the in-place goto-driven (-G2) code style.
252 -Implemented a split version of in-place goto-driven code style. This code
253 generation style is suitable for producing fast implementations of very
254 large machines. Partitioning is currently naive. In the future a
255 high-quality partitioning program will be employed. The flag for accessing
256 this feature is -Pn, where n is the number of partitions.
257 -Converted mailbox1.rl, strings2.rl and cppscan1.rl tests to support the
258 split code generation.
259 -Fixes and updates were made to the runtests script: added -c for compiling
260 only, changed the -me option to -e, and added support for testing the split
263 Ragel 5.9 - Jul 19, 2006
264 ========================
265 -Fixed a bug in the include system which caused malformed output from the
266 frontend when the include was made from a multi-line machine spec and the
267 included file ended in a single line spec (or vice versa).
268 -Static data is now const.
269 -Actions which referenced states but were not embedded caused the frontend to
271 -Manual now built with pdflatex.
272 -The manual was reorganized and expanded. Chapter sequence is now:
273 Introduction, Constructing Machines, Embedding Actions, Controlling
274 Nondeterminism and Interfacing to the Host program.
276 Ragel 5.8 - Jun 17, 2006
277 ========================
278 -The internal representation of the alphabet type has been encapsulated
279 into a class and all operations on it have been defined as C++ operators.
280 -The condition implementation now supports range transitions. This allows
281 conditions to be embedded into arbitrary machines. Conditions are still
283 -More condition embedding operators were added
284 1. Isolate the start state and embed a condition into all transitions
287 2. Embed a condition into all transitions:
288 when cond OR $when cond OR $?cond
289 3. Embed a condition into pending out transitions:
291 -Improvements were made to the determinization process to support pending out
293 -The Vim sytax file was fixed so that :> doesn't cause the match of a label.
294 -The test suite was converted to a single-file format which uses less disk
295 space than the old directory-per-test format.
297 Ragel 5.7 - May 14, 2006
298 ========================
299 -Conditions will not be embedded like actions because they involve a
300 manipulation of the state machine they are specified in. They have therefore
301 been taken out of the verbose action embedding form (using the <- compound
302 symbol). A new syntax for specifying conditions has been created:
303 m = '\n' when {i==4};
304 -Fixed a bug which prevented state machine commands like fcurs, fcall, fret,
305 etc, from being accounted for in from-state actions and to-state actions.
306 This prevented some necessary support code from being generated.
307 -Implemented condition testing in remaining code generators.
308 -Configure script now checks for gperf, which is required for building.
309 -Added support for case-insensitive literal strings (in addition to regexes).
310 A case-insensitive string is made by appending an 'i' to the literal, as in
312 -Fixed a bug which caused all or expressions inside of all regular
313 expressions to be case-insensitive. For example /[fo]o bar/ would make the
314 [fo] part case-insensitive even though no 'i' was given following the
317 Ragel 5.6 - Apr 1, 2006
318 =======================
319 -Added a left-guarded concatenation operator. This operator <: is equivalent
320 to ( expr1 $1 . expr2 >0 ). It is useful if you want to prefix a sequence
321 with a sequence of a subset of the characters it matches. For example, one
322 can consume leading whitespace before tokenizing a sequence of whitespace
323 separated words: ( ' '* <: ( ' '+ | [a-z]+ )** )
324 -Removed context embedding code, which has been dead since 5.0.
326 Ragel 5.5 - Mar 28, 2006
327 ========================
328 -Implemented a case-insensitive option for regular expressions: /get/i.
329 -If no input file is given to the ragel program it reads from standard input.
330 -The label of the start state has been changed from START to IN to save on
331 required screen space.
332 -Bug fix: \0 was not working in literal strings, due to a change that reduced
333 memory usage by concatenating components of literal strings. Token data
334 length is now passed from the scanner to the paser so that we do not need to
335 rely on null termination.
337 Ragel 5.4 - Mar 12, 2006
338 ========================
339 -Eliminated the default transition from the frontend implementation. This
340 default transition was a space-saving optimization that at best could reduce
341 the number of allocated transitions by one half. Unfortunately it
342 complicated the implementation and this stood in the way of introducing
343 conditionals. The default transition may be reintroduced in the future.
344 -Added entry-guarded concatenation. This operator :>, is syntactic sugar
345 for expr1 $0 . expr >1. This operator terminates the matching of the first
346 machine when a first character of the second machine is matched. For
347 example in any* . ';' we never leave the any* machine. If we use any* :> ';'
348 then the any* machine is terminiated upon matching the semi-colon.
349 -Added finish-guarded concatenation. This operator :>>, is syntactic sugar
350 for expr1 $0 . expr @1. This operator is like entry guarded concatenation
351 except the first machine is terminated when the second machine enters a
352 final state. This is useful for delaying the guard until a full pattern is
353 matched. For example as in '/*' any* :>> '*/'.
354 -Added strong subtraction. Where regular subtraction removes from the first
355 machine any strings that are matched by the second machine, strong
356 subtraction removes any strings from the first that contain any strings of
357 the second as a substring. Strong subtraction is syntactic sugar for
358 expr1 - ( any* expr2 any* ).
359 -Eliminated the use of priorities from the examples. Replaced with
360 subtraction, guarded concatenation and longest-match kleene star.
361 -Did some initial work on supporting conditional transitions. Far from
362 complete and very buggy. This code will only be active when conditionals are
365 Ragel 5.3 - Jan 27, 2006
366 ========================
367 -Added missing semi-colons that cause the build to fail when using older
369 -Fix for D code: if the contents of an fexec is a single word, the generated
370 code will get interpreted as a C-style cast. Adding two brackets prevents
371 this. Can now turn eliminate the "access this.;" in cppscan5 that was used to
372 get around this problem.
373 -Improved some of the tag names in the intermediate format.
374 -Added unsigned long to the list of supported alphabet types.
375 -Added ids of actions and action lists to XML intermediate format. Makes it
377 -Updated to latest Aapl package.
379 Ragel 5.2 - Jan 6, 2006
380 ========================
381 -Ragel emits an error if the target of fentry, fcall, fgoto or fnext is inside
382 a longest match operator, or if an action embedding in a longest match
383 machine uses fcall. The fcall command can still be used in pattern actions.
384 -Made improvements to the clang, rlscan, awkemu and cppscan examples.
385 -Some fixes to generated label names: they should all be prefixed with _.
386 -A fix to the Vim syntax highlighting script was made
387 -Many fixes and updates to the documentation. All important features and
388 concepts are now documented. A second chapter describing Ragel's use
391 Ragel 5.1 - Dec 22, 2005
392 ========================
393 -Fixes to the matching of section delimiters in Vim syntax file.
394 -If there is a longest match machine, the tokend var is now initialized by
395 write init. This is not necessary for correct functionality, however
396 prevents compiler warnings.
397 -The rlscan example was ported to the longest match operator and changed to
399 -Fix to the error handling in the frontend: if there are errors in the lookup
400 of names at machine generation time then do not emit anything.
401 -If not compiling the full machine in the frontend (by using -M), avoid
402 errors and segfaults caused by names that are not part of the compiled
404 -Longest match bug fix: need to init tokstart when returing from fsm calls
405 that are inside longest match actions.
406 -In Graphviz drawing, the arrow into the start state is not a real
407 transition, do not draw to-state actions on the label.
408 -A bug fix to the handling of non-tag data within an XML tag was made.
409 -Backend exit value fixed: since the parser now accepts nothing so as to
410 avoid a redundant parse error when the frontend dies, we must force an
411 error. The backend should now be properly reporting errors.
412 -The longest match machine now has it's start state set final. An LM machine
413 is in a final state when it has not matched anything, when it has matched
414 and accepted a token and is ready for another, and when it has matched a
415 token but is waiting for some lookahead before determining what to do about
416 it (similar to kleene star).
417 -Element statement removed from some tests.
418 -Entry point names are propagated to the backend and used to label the entry
419 point arrows in Graphviz output.
421 Ragel 5.0 - Dec 17, 2005
422 ========================
423 (additional details in V5 release notes)
424 -Ragel has been split into two executables: A frontend which compiles
425 machines and emits them in an XML format, and a backend which generates code
426 or a Graphviz dot file from the XML input. The purpose of this split is to
427 allow Ragel to interface with other tools by means of the XML intermediate
428 format and to reduce complexity by strictly separating the previously
429 entangled phases. The intermediate format will provide a better platform
430 inspecting compiled machines and for extending Ragel to support other host
432 -The host language interface has been reduced significantly. Ragel no longer
433 expects the machine to be implemented as a structure or class and does not
434 generate functions corresponding to initialization, execution and EOF.
435 Instead, Ragel just generates the code of these components, allowing all of
436 them to be placed in a single function if desired. The user specifies a
437 machine in the usual manner, then indicates at which place in the program
438 text the state machine code is to be generated. This is done using the write
439 statement. It is possible to specify to Ragel how it should access the
440 variables it needs (such as the current state) using the access statement.
441 -The host language embedding delimiters have been changed. Single line
442 machines start with '%%' and end at newline. Multiline machines start with
443 '%%{' and end with '}%%'. The machine name is given with the machine
444 statement at the very beginning of the specification. This purpose of this
445 change is to make it easier separate Ragel code from the host language. This
446 will ease the addition of supported host languages.
447 -The structure and class parsing which was previously able to extract a
448 machine's name has been removed since this feature is dependent on the host
449 language and inhibits the move towards a more language-independent frontend.
450 -The init, element and interface statements have been made obsolete by the
451 new host language interface and have been removed.
452 -The fexec action statement has been changed to take only the new position to
453 move to. This statement is more useful for moving backwards and reparsing
454 input than for specifying a whole new buffer entirely and has been shifted
455 to this new use. Giving it only one argument also simplifies the parsing of
456 host code embedded in a Ragel specification. This will ease the addition of
457 supported host languages.
458 -Introduced the fbreak statement, which allows one to stop processing data
459 immediately. The machine ends up in the state that the current transition
460 was to go to. The current character is not changed.
461 -Introduced the noend option for writing the execute code. This inhibits
462 checking if we have reached pe. The machine will run until it goes into the
463 error state or fbreak is hit. This allows one to parse null-terminate
464 strings without first computing the length.
465 -The execute code now breaks out of the processing loop when it moves into
466 the error state. Previously it would run until pe was hit. Breaking out
467 makes the noend option useful when an error is encountered and allows
468 user code to determine where in the input the error occured. It also
469 eliminates needlessly iterating the input buffer.
470 -Introduced the noerror, nofinal and noprefix options for writing the machine
471 data. The first two inhibit the writing of the error state and the
472 first-final state should they not be needed. The noprefix eliminates the
473 prefixing of the data items with the machine name.
474 -Support for the D language has been added. This is specified in the backend
476 -Since the new host language interface has been reduced considerably, Ragel
477 no longer needs to distinguish between C-based languages. Support for C, C++
478 and Objective-C has been folded into one option in the backend: -C
479 -The code generator has been made independent of the languages that it
480 supports by pushing the language dependent apsects down into the lower
481 levels of the code generator.
482 -Many improvements to the longest match construction were made. It is no
483 longer considered experimental. A longest match machine must appear at the
484 top level of a machine instantiation. Since it does not generate a pure
485 state machine (it may need to backtrack), it cannot be used as an operand to
487 -References to the current character and current state are now completely
488 banned in EOF actions.
490 Ragel 4.2 - Sep 16, 2005
491 ========================
492 (additional details in V4 release notes)
493 -Fixed a bug in the longest match operator. In some states it's possible that
494 we either match a token or match nothing at all. In these states we need to
495 consult the LmSwitch on error so it must be prepared to execute an error
496 handler. We therefore need to init act to this error value (which is zero).
497 We can compute if we need to do this and the code generator emits the
498 initialization only if necessary.
499 -Changed the definition of the token end of longest match actions. It now
500 points to one past the last token. This makes computing the token length
501 easier because you don't have to add one. The longest match variables token
502 start, action identifier and token end are now properly initialized in
503 generated code. They don't need to be initialized in the user's code.
504 -Implemented to-state and from-state actions. These actions are executed on
505 transitions into the state (after the in transition's actions) and on
506 transitions out of the state (before the out transition's actions). See V4
507 release notes for more information.
508 -Since there are no longer any action embedding operators that embed both on
509 transitions and on EOF, any actions that exist in both places will be there
510 because the user has explicitly done so. Presuming this case is rare, and
511 with code duplication in the hands of the user, we therefore give the EOF
512 actions their own action switch in the finish() function. This is further
513 motivated by the fact that the best solution is to do the same for to-state
514 and from-state actions in the main loop.
515 -Longest match actions can now be specified using a named action. Since a
516 word following a longest match item conflicts with the concatenation of a
517 named machine, the => symbol must come immediately before a named action.
518 -The longest match operator permits action and machine definitions in the
519 middle of a longest match construction. These are parsed as if they came
520 before the machine definition they are contained in. Permitting action and
521 machine definitions in a longest match construction allows objects to be
522 defined closer to their use.
523 -The longest match operator can now handle longest match items with no
524 action, where previously Ragel segfaulted.
525 -Updated to Aapl post 2.12.
526 -Fixed a bug in epsilon transition name lookups. After doing a name lookup
527 the result was stored in the parse tree. This is wrong because if a machine
528 is used more than once, each time it may resolve to different targets,
529 however it will be stored in the same place. We now store name resolutions
530 in a separated data structure so that each walk of a parse tree uses the
531 name resolved during the corresponding walk in the name lookup pass.
532 -The operators used to embed context and actions into states have been
533 modified. The V4 release notes contain the full details.
534 -Added zlen builtin machine to represent the zero length machine. Eventually
535 the name "null" will be phased out in favour of zlen because it is unclear
536 whether null matches the zero length string or if it does not match any
537 string at all (as does the empty builtin).
538 -Added verbose versions of action, context and priority embedding. See the V4
539 release notes for the full details. A small example:
540 machine <- all exec { foo(); } <- final eof act1
541 -Bugfix for machines with epsilon ops, but no join operations. I had
542 wrongfully assumed that because epsilon ops can only increase connectivity,
543 that no states are ever merged and therefore a call to fillInStates() is not
544 necessary. In reality, epsilon transitions within one machine can induce the
545 merging of states. In the following, state 2 follows two paths on 'i':
546 main := 'h' -> i 'i h' i: 'i';
547 -Changed the license of the guide from a custom "do not propagate modified
548 versions of this document" license to the GPL.
550 Ragel 4.1 - Jun 26, 2005
551 ========================
552 (additional details in V4 release notes)
553 -A bug in include processing was fixed. Surrounding code in an include file
554 was being passed through to the output when it should be ignored. Includes
555 are only for including portions of another machine into he current. This
556 went unnoticed because all tested includes were wrapped in #ifndef ...
557 #endif directives and so did not affect the compilation of the file making
559 -Fixes were made to Vim syntax highlighting file.
560 -Duplicate actions are now removed from action lists.
561 -The character-level negation operator ^ was added. This operator produces a
562 machine that matches single characters that are not matched by the machine
563 it is applied to. This unary prefix operator has the same precedence level
565 -The use of + to specify the a positive literal number was discontinued.
566 -The parser now assigns the subtraction operator a higher precedence than
567 the negation of literal number.
569 Ragel 4.0 - May 26, 2005
570 ========================
571 (additional details in V4 release notes)
572 -Operators now strictly embed into a machine either on a specific class of
573 characters or on EOF, but never both. This gives a cleaner association
574 between the operators and the physical state machine entitites they operate
575 on. This change is made up of several parts:
576 1. '%' operator embeds only into leaving characters.
577 2. All global and local error operators only embed on error character
578 transitions, their action will not be triggerend on EOF in non-final
580 3. EOF action embedding operators have been added for all classes of states
581 to make up for functionality removed from other operators. These are
583 4. Start transition operator '>' no longer implicitly embeds into leaving
584 transtions when start state is final.
585 -Ragel now emits warnings about the improper use of statements and values in
586 action code that is embedded as an EOF action. Warnings are emitted for fpc,
587 fc, fexec, fbuf and fblen.
588 -Added a longest match construction operator |* machine opt-action; ... *|.
589 This is for repetition where an ability to revert to a shorter, previously
590 matched item is required. This is the same behaviour as flex and re2c. The
591 longest match operator is not a pure FSM construction, it introduces
592 transitions that implicitly hold the current character or reset execution to
593 a previous location in the input. Use of this operator requires the caller
594 of the machine to occasionally hold onto data after a call to the exectute
595 routine. Use of machines generated with this operator as the input to other
596 operators may have undefined results. See examples/cppscan for an example.
597 This is very experimental code.
598 -Action ids are only assigned to actions that are referenced in the final
599 constructed machine, preventing gaps in the action id sequence. Previously
600 an action id was assigned if the action was referenced during parsing.
601 -Machine specifications now begin with %% and are followed with an optional
602 name and either a single Ragel statement or a sequence of statements
604 -Ragel no longer generates the FSM's structure or class. It is up to the user
605 to declare the structure and to give it a variable named curs of type
606 integer. If the machine uses the call stack the user must also declare a
607 array of integers named stack and an integer variable named top.
608 -In the case of Objective-C, Ragel no longer generates the interface or
609 implementation directives, allowing the user to declare additional methods.
610 -If a machine specification does not have a name then Ragel tries to find a
611 name for it by first checking if the specification is inside a struct, class
612 or interface. If it is not then it uses the name of the previous machine
613 specification. If still no name is found then this is an error.
614 -Fsm specifications now persist in memory and statements accumulate.
615 -Ragel now has an include statement for including the statements of a machine
616 spec in another file (perhaps because it is the corresponding header file).
617 The include statement can also be used to draw in the statements of another
618 fsm spec in the current file.
619 -The fstack statement is now obsolete and has been removed.
620 -A new statement, simply 'interface;', indicates that ragel should generate
621 the machine's interface. If Ragel sees the main machine it generates the
622 code sections of the machine. Previously, the header portion was generated
623 if the (now removed) struct statement was found and code was generated if
624 any machine definition was found.
625 -Fixed a bug in the resolution of fsm name references in actions. The name
626 resolution code did not recurse into inline code items with children
627 (fgoto*, fcall*, fnext*, and fexec), causing a segfault at code generation
629 -Cleaned up the code generators. FsmCodeGen was made into a virtual base
630 class allowing for the language/output-style specific classes to inherit
631 both a language specific and style-specific base class while retaining only
632 one copy of FsmCodeGen. Language specific output can now be moved into the
633 language specific code generators, requiring less duplication of code in the
634 language/output-style specific leaf classes.
635 -Fixed bugs in fcall* implementation of IpgGoto code generation.
636 -If the element type has not been defined Ragel now uses a constant version
637 of the alphtype, not the exact alphtype. In most cases the data pointer of
638 the execute routine should be const. A non-const element type can still be
639 defined with the element statement.
640 -The fc special value now uses getkey for retrieving the current char rather
641 than *_p, which is wrong if the element type is a structure.
642 -User guide converted to TeX and updated for new 4.0 syntax and semantics.
644 Ragel 3.7 - Oct 31, 2004
645 ========================
646 -Bug fix: unreferenced machine instantiations causing segfault due to name
647 tree and parse tree walk becomming out of syncronization.
648 -Rewrote representation of inline code blocks using a tree data structure.
649 This allows special keywords such as fbuf to be used as the operatands of
651 -Documentation updates.
652 -When deciding whether or not to generate machine instantiations, search the
653 entire name tree beneath the instantiation for references, not just the
655 -Removed stray ';' in keller2.rl
656 -Added fexec for restarting the machine with new buffer data (state stays the
657 same), fbuf for retrieving the the start of the buf, and fblen for
658 retrieving the orig buffer length.
659 -Implemented test/cppscan2 using fexec. This allows token emitting and restart
660 to stay inside the execute routine, instead of leaving and re-entering on
662 -Changed examples/cppscan to use fexec and thereby go much faster.
663 -Implemented flex and re2c versions of examples/cppscan. Ragel version
664 goes faster than flex version but not as fast as re2c version.
665 -Merged in Objective-C patch from Erich Ocean.
666 -Turned off syncing with stdio in C++ tests to make them go faster.
667 -Renamed C++ code generaion classes with the Cpp Prefix instead of CC to make
669 -In the finish function emit fbuf as 0 cast to a pointer to the element type
670 so it's type is not interpreted as an integer.
671 -The number -128 underflows char alphabets on some architectures. Removed
673 -Disabled the keller2 test because it causes problems on many architectures
674 due to its large size and compilation requirements.
676 Ragel 3.6 - Jul 10, 2004
677 ========================
678 -Many documentation updates.
679 -When resolving names, return a set of values so that a reference in an
680 action block that is embedded more than once won't report distinct entry
681 points that are actually the same.
682 -Implemented flat tables. Stores a linear array of indicies into the
683 transition array and only a low and high key value. Faster than binary
684 searching for keys but not usable for large alphabets.
685 -Fixed bug in deleting of transitions leftover from converstion from bst to
686 list implementation of transitions. Other code cleanup.
687 -In table based output calculate the cost of using an index. Don't use if
689 -Changed fstate() value available in init and action code to to fentry() to
690 reflect the fact that the values returned are intended to be used as targets
691 in fgoto, fnext and fcall statements. The returned state is not a unique
692 state representing the label. There can be any number of states representing
694 -Added keller2 test, C++ scanning tests and C++ scanning example.
695 -In table based output split up transitions into targets and actions. This
696 allows actions to be omitted.
697 -Broke the components of the state array into separate arrays. Requires
698 adding some fields where they could previously be omitted, however allows
699 finer grained control over the sizes of items and an overal size reduction.
700 Also means that state numbers are not an offset into the state array but
701 instead a sequence of numbers, meaning the context array does not have any
703 -Action lists and transition also have their types chosen to be the smallest
704 possible for accomodating the contained values.
705 -Changed curs state stored in fsm struct from _cs to curs. Keep fsm->curs ==
706 -1 while in machine. Added tests curs1 and curs2.
707 -Implemented the notion of context. Context can be embedded in states using
708 >:, $:, @: and %: operators. These embed a named context into start states,
709 all states, non-start/non-final and final states. If the context is declared
710 using a context statment
712 then the context can be quered for any state using fsm_name_ctx_name(state)
713 in C code and fsm_name::ctx_name(state) in C++ code. This feature makes it
714 possible to determine what "part" of the machine is currently active.
715 -Fixed crash on machine generation of graphs with no final state. If there
716 is no reference to a final state in a join operation, don't generate one.
717 -Updated Vim sytax: added labels to inline code, added various C++ keywords.
718 Don't highlight name separations as labels. Added switch labels, improved
719 alphtype, element and getkey.
720 -Fixed line info in error reporting of bad epsilon trans.
721 -Fixed fstate() for tab code gen.
722 -Removed references to malloc.h.
724 Ragel 3.5 - May 29, 2004
725 ========================
726 -When parse errors occur, the partially generated output file is deleted and
727 an non-zero exit status is returned.
728 -Updated Vim syntax file.
729 -Implemented the setting of the element type that is passed to the execute
730 routine as well as method for specifying how ragel should retrive the key
731 from the element type. This lets ragel process arbitrary structures inside
732 of which is the key that is parsed.
733 element struct Element;
734 getkey fpc->character;
735 -The current state is now implemented with an int across all machines. This
736 simplifies working with current state variables. For example this allows a
737 call stack to be implemented in user code.
738 -Implemented a method for retrieving the current state, the target state, and
740 fcurs -retrieve the current state
741 ftargs -retrieve the target state
742 fstate(name) -retrieve a named state.
743 -Implemented a mechanism for jumping to and calling to a state stored in a
745 fgoto *<expr>; -goto the state returned by the C/C++ expression.
746 fcall *<expr>; -call the state returned by the C/C++ expression.
747 -Implemented a mechanism for specifying the next state without immediately
748 transfering control there (any code following statement is executed).
749 fnext label; -set the state pointed to by label as the next state.
750 fnext *<expr>; -set the state returned by the C/C++ expression as the
752 -Action references are determined from the final machine instead of during
753 the parse tree walk. Some actions can be referenced in the parse tree but not
754 show up in the final machine. Machine analysis is now done based on this new
756 -Named state lookup now employs a breadth-first search in the lookup and
757 allows the user to fully qualify names, making it possible to specify
758 jumps/calls into parts of the machine deep in the name hierarchy. Each part
759 of name (separated by ::) employs a breadth first search from it's starting
761 -Name references now must always refer to a single state. Since references to
762 multiple states is not normally intended, it no longer happens
763 automatically. This frees the programmer from thinking about whether or not
764 a state reference is unique. It also avoids the added complexity of
765 determining when to merge the targets of multiple references. The effect of
766 references to multiple states can be explicitly created using the join
767 operator and epsilon transitions.
768 -M option was split into -S and -M. -S specifies the machine spec to generate
769 for graphviz output and dumping. -M specifies the machine definition or
771 -Machine function parameters are now prefixed with and underscore to
772 avoid the hiding of class members.
774 Ragel 3.4 - May 8, 2004
775 =======================
776 -Added the longest match kleene star operator **, which is synonymous
777 with ( ( <machine> ) $0 %1 ) *.
778 -Epsilon operators distinguish between leaving transitions (going to an
779 another expression in a comma separated list) and non-leaving transitions.
780 Leaving actions and priorities are appropriately transferred.
781 -Relative priority of following ops changed to:
785 If label is done first then the isolation of the start state in > operators
786 will cause the label to point to the old start state that doesn't have the
788 -Merged >! and >~, @! and @~, %! and %~, and $! and $~ operators to have one
789 set of global error action operators (>!, @!, %! and $!) that are invoked on
790 error by unexpected characters as well as by unexepected EOF.
791 -Added the fpc keyword for use in action code. This is a pointer to the
792 current character. *fpc == fc. If an action is invoked on EOF then fpc == 0.
793 -Added >^, @^, %^, and $^ local error operators. Global error operators (>!,
794 @!, $!, and %!) cause actions to be invoked if the final machine fails.
795 Local error actions cause actions to be invoked if if the current machine
797 -Changed error operators to mean embed global/local error actions in:
798 >! and !^ -the start state.
799 @! and @^ -states that are not the start state and are not final.
800 %! and %^ -final states.
801 $! and $^ -all states.
802 -Added >@! which is synonymous >! then @!
803 -Added >@^ which is synonymous >^ then @^
804 -Added @%! which is synonymous @! then %!
805 -Added @%^ which is synonymous >^ then @^
806 -FsmGraph representation of transition lists was changed from a mapping of
807 alphabet key -> transition objects using a BST to simply a list of
808 transition objects. Since the transitions are no longer divided by
809 single/range, the fast finding of transition objects by key is no longer
810 required functionality and can be eliminated. This new implementation uses
811 the same amount of memory however causes less allocations. It also make more
812 sense for supporting error transitions with actions. Previously an error
813 transition was represented by a null value in the BST.
814 -Regular expression ranges are checked to ensure that lower <= upper.
815 -Added printf-like example.
816 -Added atoi2, erract2, and gotcallret to the test suite.
817 -Improved build test to support make -jN and simplified the compiling and
820 Ragel 3.3 - Mar 7, 2004
821 =======================
822 -Portability bug fixes were made. Minimum and maximum integer values are
823 now taken from the system. An alignment problem on 64bit systems
826 Ragel 3.2 - Feb 28, 2004
827 ========================
828 -Added a Vim syntax file.
829 -Eliminated length var from generated execute code in favour of an end
830 pointer. Using length requires two variables be read and written. Using an
831 end pointer requires one variable read and written and one read. Results in
832 more optimizable code.
833 -Minimization is now on by default.
834 -States are ordered in output by depth first search.
835 -Bug in minimization fixed. States were not being distinguished based on
837 -Added null and empty builtin machines.
838 -Added EOF error action operators. These are >~, >@, $~, and %~. EOF error
839 operators embed actions to take if the EOF is seen and interpreted as an
840 error. The operators correspond to the following states:
842 -any state with a transition to a final state
843 -any state with a transiion out
845 -Fixed bug in generation of unreference machine vars using -M. Unreferenced
846 vars don't have a name tree built underneath when starting from
847 instantiations. Need to instead build the name tree starting at the var.
848 -Calls, returns, holds and references to fc in out action code are now
849 handled for ipgoto output.
850 -Only actions referenced by an instantiated machine expression are put into
851 the action index and written out.
852 -Added rlscan, an example that lexes Ragel input.
854 Ragel 3.1 - Feb 18, 2004
855 ========================
856 -Duplicates in OR literals are removed and no longer cause an assertion
858 -Duplicate entry points used in goto and call statements are made into
859 deterministic entry points.
860 -Base FsmGraph code moved from aapl into ragel, as an increasing amount
861 of specialization is required. Too much time was spent attempting to
862 keep it as a general purpose template.
863 -FsmGraph code de-templatized and heirarchy squashed to a single class.
864 -Single transitions taken out of FsmGraph code. In the machine construction
865 stage, transitions are now implemented only with ranges and default
866 transtions. This reduces memory consumption, simplifies code and prevents
867 covered transitions. However it requires the automated selection of single
868 transitions to keep goto-driven code lean.
869 -Machine reduction completely rewritten to be in-place. As duplicate
870 transitions and actions are found and the machine is converted to a format
871 suitable for writing as C code or as GraphViz input, the memory allocated
872 for states and transitions is reused, instead of newly allocated.
873 -New reduction code consolodates ranges, selects a default transition, and
874 selects single transitions with the goal of joining ranges that are split by
875 any number of single characters.
876 -Line directive changed from "# <num> <file>" to the more common format
877 "#line <num> <file>".
878 -Operator :! changed to @!. This should have happened in last release.
879 -Added params example.
881 Ragel 3.0 - Jan 22, 2004
882 ========================
883 -Ragel now parses the contents of struct statements and action code.
884 -The keyword fc replaces the use of *p to reference the current character in
886 -Machine instantiations other than main are allowed.
887 -Call, jump and return statements are now available in action code. This
888 facility makes it possible to jump to an error handling machine, call a
889 sub-machine for parsing a field or to follow paths through a machine as
890 determined by arbitrary C code.
891 -Added labels to the language. Labels can be used anywhere in a machine
892 expression to define an entry point. Also references to machine definitions
893 cause the implicit creation of a label.
894 -Added epsilon transitions to the language. Epsilon operators may reference
895 labels in the current name scope resolved when join operators are evaluated
896 and at the root of the expression tree of machine assignment/instantiation.
897 -Added the comma operator, which joins machines together without drawing any
898 transitions between them. This operator is useful in combination with
899 labels, the epsilon operator and user code transitions for defining machines
900 using the named state and transition list paradigm. It is also useful for
901 invoking transitions based on some analysis of the input or on the
903 -Added >!, :!, $!, %! operators for specifying actions to take should the
904 machine fail. These operators embed actions to execute if the machine
907 -any state with a transition to a final state
908 -any state with a transiion out
910 The general rule is that if an action embedding operator embeds an action
911 into a set of transitions T, then the error-counterpart with a ! embeds an
912 action into the error transition taken when any transition T is a candidate,
913 but does not match the input.
914 -The finishing augmentation operator ':' has been changed to '@'. This
915 frees the ':' symbol for machine labels and avoids hacks to the parser to
916 allow the use of ':' for both labels and finishing augmentations. The best
917 hack required that label names be distinct from machine definition names as
918 in main := word : word; This restriction is not good because labels are
919 local to the machine that they are used in whereas machine names are global
920 entities. Label name choices should not be restricted by the set of names
921 that are in use for machines.
922 -Named priority syntax now requires parenthesis surrounding the name and
923 value pair. This avoids grammar ambiguities now that the ',' operator has
924 been introduced and makes it more clear that the name and value are an
926 -Backslashes are escaped in line directive paths.
928 Ragel 2.2 - Oct 6, 2003
929 =======================
930 -Added {n}, {,n}, {n,} {n,m} repetition operators.
931 <expr> {n} -- exactly n repetitions
932 <expr> {,n} -- zero to n repetitions
933 <expr> {n,} -- n or more repetitions
934 <expr> {n,m} -- n to m repetitions
935 -Bug in binary search table in Aapl fixed. Fixes crashing on machines that
936 add to action tables that are implicitly shared among transitions.
937 -Tests using obsolete minimization algorithms are no longer built and run by
939 -Added atoi and concurrent from examples to the test suite.
941 Ragel 2.1 - Sep 22, 2003
942 ========================
943 -Bug in priority comparison code fixed. Segfaulted on some input with many
945 -Added two new examples.
947 Ragel 2.0 - Sep 7, 2003
948 =======================
949 -Optional (?), One or More (+) and Kleene Star (*) operators changed from
950 prefix to postfix. Rationale is that postfix version is far more common in
951 regular expression implementations and will be more readily understood.
952 -All priority values attached to transitions are now accompanied by a name.
953 Transitions no longer have default priority values of zero assigned
954 to them. Only transitions that have different priority values assigned
955 to the same name influence the NFA-DFA conversion. This scheme reduces
956 side-effects of priorities.
957 -Removed the %! statement for unsetting pending out priorities. With
958 named priorities, it is not necessary to clear the priorities of a
959 machine with $0 %! because non-colliding names can be used to avoid
961 -Removed the clear keyword, which was for removing actions from a machine.
962 Not required functionality and it is non-intuitive to have a language
963 feature that undoes previous definitions.
964 -Removed the ^ modifier to repetition and concatenation operators. This
965 undocumented feature prevented out transitions and out priorities from being
966 transfered from final states to transitions leaving machines. Not required
967 functionality and complicates the language unnecessarily.
968 -Keyword 'func' changed to 'action' as a part of the phasing out of the term
969 'function' in favour of 'action'. Rationale is that the term 'function'
970 implies that the code is called like a C function, which is not necessarily
971 the case. The term 'action' is far more common in state machine compiler
973 -Added the instantiation statement, which looks like a standard variable
974 assignment except := is used instead of =. Instantiations go into the
975 same graph dictionary as definitions. In the the future, instantiations
976 will be used as the target for gotos and calls in action code.
977 -The main graph should now be explicitly instantiated. If it is not,
979 -Or literal basic machines ([] outside of regular expressions) now support
981 -C and C++ interfaces lowercased. In the C interface an underscore now
982 separates the fsm machine and the function name. Rationale is that lowercased
983 library and generated routines are more common.
985 int fsm_init( struct clang *fsm );
986 int fsm_execute( struct clang *fsm, char *data, int dlen );
987 int fsm_finish( struct clang *fsm );
990 int fsm::execute( char *data, int dlen );
992 -Init, execute and finish all return -1 if the machine is in the error state
993 and can never accept, 0 if the machine is in a non-accepting state that has a
994 path to a final state and 1 if the machine is in an accepting state.
995 -Accept routine eliminated. Determining whether or not the machine accepts is
996 done by examining the return value of the finish routine.
997 -In C output, fsm structure is no longer a typedef, so referencing requires
998 the struct keyword. This is to stay in line with C language conventions.
999 -In C++ output, constructor is no longer written by ragel. As a consequence,
1000 init routine is not called automatically. Allows constructor to be supplied
1001 by user as well as the return value of init to be examined without calling it
1003 -Static start state and private structures are taken out of C++ classes.
1005 Ragel 1.5.4 - Jul 14, 2003
1006 ==========================
1007 -Workaround for building with bison 1.875, which produces an
1008 optimization that doesn't build with newer version gcc.
1010 Ragel 1.5.3 - Jul 10, 2003
1011 ==========================
1012 -Fixed building with versions of flex that recognize YY_NO_UNPUT.
1013 -Fixed version numbers in ragel.spec file.
1015 Ragel 1.5.2 - Jul 7, 2003
1016 =========================
1017 -Transition actions and out actions displayed in the graphviz output.
1018 -Transitions on negative numbers handled in graphviz output.
1019 -Warning generated when using bison 1.875 now squashed.
1021 Ragel 1.5.1 - Jun 21, 2003
1022 ==========================
1023 -Bugs fixed: Don't delete the output objects when writing to standard out.
1024 Copy mem into parser buffer with memcpy, not strcpy. Fixes buffer mem errror.
1025 -Fixes for compiling with Sun WorkShop 6 compilers.
1027 Ragel 1.5.0 - Jun 10, 2003
1028 ==========================
1029 -Line directives written to the output so that errors in the action code
1030 are properly reported in the ragel input file.
1031 -Simple graphviz dot file output format is supported. Shows states and
1032 transitions. Does not yet show actions.
1033 -Options -p and -f dropped in favour of -d output format.
1034 -Added option -M for specifying the machine to dump with -d or the graph to
1036 -Error recovery implemented.
1037 -Proper line and column number tracking implemented in the scanner.
1038 -All action/function code is now embedded in the main Execute routine. Avoids
1039 duplication of action code in the Finish routine and the need to call
1040 ExecFuncs which resulted in huge code bloat. Will also allow actions to
1041 modify cs when fsm goto, call and return is supported in action code.
1042 -Fsm spec can have no statements, nothing will be generated.
1043 -Bug fix: Don't accept ] as the opening of a .-. range a reg exp.
1044 -Regular expression or set ranges (ie /[0-9]/) are now handled by the parser
1045 and consequently must be well-formed. The following now generates a parser
1046 error: /[+-]/ and must be rewritten as /[+\-]/. Also fixes a bug whereby ]
1047 might be accepted as the opening of a .-. range causing /[0-9]-[0-9]/ to
1049 -\v, \f, and \r are now treated as whitespace in an fsm spec.
1051 Ragel 1.4.1 - Nov 19, 2002
1052 ==========================
1053 -Compile fixes. The last release (integer alphabets) was so exciting
1054 that usual portability checks got bypassed.
1056 Ragel 1.4.0 - Nov 19, 2002
1057 ==========================
1058 -Arbitrary integer alphabets are now fully supported! A new language
1060 'alphtype <type>' added for specifying the type of the alphabet. Default
1061 is 'char'. Possible alphabet types are:
1062 char, unsigned char, short, unsigned short, int, unsigned int
1063 -Literal machines specified in decimal format can now be negative when the
1064 alphabet is a signed type.
1065 -Literal machines (strings, decimal and hex) have their values checked for
1066 overflow/underflow against the size of the alphabet type.
1067 -Table driven and goto driven output redesigned to support ranges. Table
1068 driven uses a binary search for locating single characters and ranges. Goto
1069 driven uses a switch statement for single characters and nested if blocks for
1071 -Switch driven output removed due to a lack of consistent advantages. Most of
1072 the time the switch driven FSM is of no use because the goto FSM makes
1073 smaller and faster code. Under certain circumstances it can produce smaller
1074 code than a goto driven fsm and be almost as fast, but some sporadic case
1075 does not warrant maintaining it.
1076 -Many warnings changed to errors.
1077 -Added option -p for printing the final fsm before minimization. This lets
1078 priorities be seen. Priorties are all reset to 0 before minimization. The
1079 exiting option -f prints the final fsm after minimization.
1080 -Fixed a bug in the clang test and example that resulted in redundant actions
1083 Ragel 1.3.4 - Nov 6, 2002
1084 =========================
1085 -Fixes to Chapter 1 of the guide.
1086 -Brought back the examples and made them current.
1087 -MSVC is no longer supported for compiling windows binaries because its
1088 support for the C++ standard is frustratingly inadequate, it will cost money
1089 to upgrade if it ever gets better, and MinGW is a much better alternative.
1090 -The build system now supports the --host= option for building ragel
1091 for another system (used for cross compiling a windows binary with MinGW).
1092 -Various design changes and fixes towards the goal of arbitrary integer
1093 alphabets and the handling of larger state machines were made.
1094 -The new shared vector class is now used for action lists in transitions and
1095 states to reduce memory allocations.
1096 -An avl tree is now used for the reduction of transitions and functions of an
1097 fsm graph before making the final machine. The tree allows better scalability
1098 and performance by not requiring consecutively larger heap allocations.
1099 -Final stages in the separation of fsm graph code from action embedding and
1100 priority assignment is complete. Makes the base graph leaner and easier to reuse
1101 in other projects (like Keller).
1103 Ragel 1.3.3 - Oct 22, 2002
1104 ==========================
1105 -More diagrams were added to section 1.7.1 of the user guide.
1106 -FSM Graph code was reworked to spearate the regex/nfa/minimizaion graph
1107 algorithms from the manipulation of state and transition properties.
1108 -An rpm spec file from Cris Bailiff was added. This allows an rpm for ragel
1109 to be built with the command 'rpm -ta ragel-x.x.x.tar.gz'
1110 -Fixes to the build system and corresponding doc updates in the README.
1111 -Removed autil and included the one needed source file directly in the top
1112 level ragel directory.
1113 -Fixed a bug that nullified the 20 times speedup in large compilations
1114 claimed by the last version.
1115 -Removed awk from the doc build (it was added with the last release -- though
1116 not mentioned in the changelog).
1117 -Install of man page was moved to the doc dir. The install also installs the
1118 user guide to $(PREFIX)/share/doc/ragel/
1120 Ragel 1.3.2 - Oct 16, 2002
1121 ==========================
1122 -Added option -v (or --version) to show version information.
1123 -The subtract operator no longer removes transition data from the machine
1124 being subtracted. This is left up to the user for the purpose of making it
1125 possible to transfer transitions using subtract and also for speeding up the
1126 subtract routine. Note that it is possible to explicitly clear transition
1127 data before a doing a subtract.
1128 -Rather severe typo bug fixed. Bug was related to transitions with higher
1129 priorities taking precedence. A wrong ptr was being returned. It appears to
1130 have worked most of the time becuase the old ptr was deleted and the new one
1131 allocated immediatly after so the old ptr often pointed to the same space.
1133 -Bug in the removing of dead end paths was fixed. If the start state
1134 has in transitions then those paths were not followed when finding states to
1135 keep. Would result in non-dead end states being removed from the graph.
1136 -In lists and in ranges are no longer maintained as a bst with the key as the
1137 alphabet character and the value as a list of transitions coming in on that
1138 char. There is one list for each of inList, inRange and inDefault. Now that
1139 the required functionality of the graph is well known it is safe to remove
1140 these lists to gain in speed and footprint. They shouldn't be needed.
1141 -IsolateStartState() runs on modification of start data only if the start
1142 state is not already isolated, which is now possible with the new in list
1144 -Concat, Or and Star operators now use an approximation to
1145 removeUnreachableStates that does not require a traversal of the entire
1146 graph. This combined with an 'on-the-fly' management of final bits and final
1147 state status results is a dramatic speed increase when compiling machines
1148 that use those operators heavily. The strings2 test goes 20 times faster.
1149 -Before the final minimization, after all fsm operations are complete,
1150 priority data is reset which enables better minimization in cases where
1151 priorities would otherwise separate similar states.
1153 Ragel 1.3.1 - Oct 2, 2002
1154 =========================
1155 -Range transitions are now used to implement machines made with /[a-z]/ and
1156 the .. operator as well as most of the builtin machines. The ranges are not
1157 yet reflected in the output code, they are expanded as if they came from the
1158 regular single transitions. This is one step closer to arbitrary integer
1160 -The builtin machine 'any' was added. It is equiv to the builtin extend,
1161 matching any characters.
1162 -The builtin machine 'cntrl' now includes newline.
1163 -The builtin machine 'space' now includes newline.
1164 -The builtin machine 'ascii' is now the range 0-127, not all characters.
1165 -A man page was written.
1166 -A proper user guide was started. Chapter 1: Specifying Ragel Programs
1167 was written. It even has some diagrams :)
1169 Ragel 1.3.0 - Sep 4, 2002
1170 =========================
1171 -NULL keyword no longer used in table output.
1172 -Though not yet in use, underlying graph structure changed to support range
1173 transitions. As a result, most of the code that walks transition lists is now
1174 implemented with an iterator that hides the complexity of the transition
1175 lists and ranges. Range transitions will be used to implement /[a-z]/ style
1176 machines and machines made with the .. operator. Previously a single
1177 transition would be used for each char in the range, which is very costly.
1178 Ranges eliminate much of the space complexity and allow for the .. operator
1179 to be used with very large (integer) alphabets.
1180 -New minimization similar to Hopcroft's alg. It does not require n^2 space and
1181 runs close to O(n*log(n)) (an exact analysis of the alg is very hard). It is
1182 much better than the stable and approx minimization and obsoletes them both.
1183 An exact implementation of Hopcroft's alg is desirable but not possible
1184 because the ragel implementation does not assume a finite alphabet, which
1185 Hopcroft's requires. Ragel will support arbitrary integer alphabets which
1186 must be treated as an infinite set for implementation considerations.
1187 -New option -m using above described minimization to replace all previous
1188 minimization options. Old options sill work but are obsolete and not
1190 -Bug fixed in goto style output. The error exit set the current state to 0,
1191 which is actually a valid state. If the machine was entered again it would go
1192 into the first state, very wrong. If the first state happened to be final then
1193 an immediate finish would accept when in fact it should fail.
1194 -Slightly better fsm minimization now capable due to clearing of the
1195 transition ordering numbers just prior to minimization.
1197 Ragel 1.2.2 - May 25, 2002
1198 ==========================
1199 -Configuration option --prefix now works when installing.
1200 -cc file extension changed to cpp for better portability.
1201 -Unlink of output file upon error no longer happens, removes dependency on
1202 unlink system command.
1203 -All multiline strings removed: not standard c++.
1204 -Awk build dependency removed.
1205 -MSVC 6.0 added to the list of supported compilers (with some tweaking of
1206 bison and flex output).
1208 Ragel 1.2.1 - May 13, 2002
1209 ==========================
1210 -Automatic dependencies were fixed, they were not working correctly.
1211 -Updated AUTHORS file to reflect contributors.
1212 -Code is more C++ standards compliant: compiles with g++ 3.0
1213 -Fixed bugs that only showed up in g++ 3.0
1214 -Latest (unreleased) Aapl.
1215 -Configuration script bails out if bison++ is installed. Ragel will not
1216 compile with bison++ because it is coded in c++ and bison++ automatically
1217 generates a c++ parser. Ragel uses a c-style bison parser.
1219 Ragel 1.2.0 - May 3, 2002
1220 =========================
1221 -Underlying graph structure now supports default transitions. The result is
1222 that a transition does not need to be made for each char of the alphabet
1223 when making 'extend' or '/./' machines. Ragel compiles machines that
1224 use the aforementioned primitives WAY faster.
1225 -The ugly hacks needed to pick default transitions now go away due to
1226 the graph supporting default transitions directly.
1227 -If -e is given, but minimization is not turned on, print a warning.
1228 -Makefiles use automatic dependencies.
1230 Ragel 1.1.0 - Apr 15, 2002
1231 ==========================
1232 -Added goto fsm: much faster than any other fsm style.
1233 -Default operator (if two machines are side by side with no operator
1234 between them) is concatenation. First showed up in 1.0.4.
1235 -The fsm machine no longer auotmatically builds the flat table for
1236 transition indicies. Instead it keeps the key,ptr pair. In tabcodegen
1237 the flat table is produced. This way very large alphabets with sparse
1238 transitions will not consume large amounts of mem. This is also in prep
1239 for fsm graph getting a default transition.
1240 -Generated code contains a statement explicitly stating that ragel fsms
1241 are NOT covered by the GPL. Technically, Ragel copies part of itself
1242 to the output to make the generic fsm execution routine (for table driven
1243 fsms only) and so the output could be considered under the GPL. But this
1244 code is very trivial and could easlily be rewritten. The actual fsm data
1245 is subject to the copyright of the source. To promote the use of Ragel,
1246 a special exception is made for the part of the output copied from Ragel:
1247 it may be used without restriction.
1248 -Much more elegant code generation scheme is employed. Code generation
1249 class members need only put the 'codegen' keyword after their 'void' type
1250 in order to be automatically registerd to handle macros of the same name.
1251 An awk script recognises this keyword and generates an appropriate driver.
1252 -Ragel gets a test suite.
1253 -Postfunc and prefunc go away because they are not supported by non
1254 loop-driven fsms (goto, switch) and present duplicate functionality.
1255 Universal funcs can be implemented by using $ operator.
1256 -Automatic dependencies used in build system, no more make depend target.
1257 -Code generation section in docs.
1258 -Uses the latests aapl.
1260 Ragel 1.0.5 - Mar 3, 2002
1261 =========================
1262 -Bugfix in SetErrorState that caused an assertion failure when compiling
1263 simple machines that did not have full transition tables (and thus did
1264 not show up on any example machines). Assertion failure did not occur
1265 when using the switch statement code as ragel does not call SetErrorState
1267 -Fixed some missing includes, now compiles on redhat.
1268 -Moved the FsmMachTrans Compare class out of FsmMachTrans. Some compilers
1269 don't deal with nested classes in templates too well.
1270 -Removed old unused BASEREF in fsmgraph and ragel now compiles using
1271 egcs-2.91.66 and presumably SUNWspro. The baseref is no longer needed
1272 because states do not support being elements in multiple lists. I would
1273 rather be able to support more compilers than have this feature.
1274 -Started a README with compilation notes. Started an AUTHORS file.
1275 -Started the user documentation. Describes basic machines and operators.
1277 Ragel 1.0.4 - Mar 1, 2002
1278 =========================
1279 -Ported to the version of Aapl just after 2.2.0 release. See
1280 http://www.ragel.ca/aapl/ for details on aapl.
1281 -Fixed a bug in the clang example: the newline machine was not stared.
1282 -Added explanations to the clang and mailbox examples. This should
1283 help people that want to learn the lanuage as the manual is far from
1286 Ragel 1.0.3 - Feb 2, 2002
1287 =========================
1288 -Added aapl to the ragel tree. No longer requires you to download
1289 and build aapl separately. Should avoid discouraging impatient users
1290 from compiling ragel.
1291 -Added the examples to the ragel tree.
1292 -Added configure script checks for bison and flex.
1293 -Fixed makefile so as not to die with newer versions of bison that
1294 write the header of the parser to a .hh file.
1295 -Started ChangeLog file.
1297 Ragel 1.0.2 - Jan 30, 2002
1298 ==========================
1299 -Bug fix in calculating highIndex for table based code. Was using
1300 the length of out tranisition table rather than the value at the
1302 -If high/low index are at the limits, output a define in their place,
1303 not the high/low values themselves so as not to cause compiler warnings.
1304 -If the resulting machines don't have any indicies or functions, then
1305 omit the empty unrefereced static arrays so as not to cause compiler
1306 warnings about unused static vars.
1307 -Fixed variable sized indicies support. The header cannot have any
1308 reference to INDEX_TYPE as that info is not known at the time the header
1309 data is written. Forces us to use a void * for pointers to indicies. In
1310 the c++ versions we are forced to make much of the data non-member
1311 static data in the code portion for the same reason.
1313 Ragel 1.0.1 - Jan 28, 2002
1314 ==========================
1315 -Exe name change from reglang to ragel.
1316 -Added ftabcodegen output code style which uses a table for states and
1317 transitions but uses a switch statement for the function execution.
1318 -Reformatted options in usage dump to look better.
1319 -Support escape sequences in [] sections of regular expressions.
1321 Ragel 1.0 - Jan 25, 2002
1322 ========================