Return to WHATSNEW CVS log

Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / php / ext / ereg / regex

Annotation of embedaddon/php/ext/ereg/regex/WHATSNEW, revision 1.1.1.1

1.1       misho       1: New in alpha3.4:  The complex bug alluded to below has been fixed (in a
                      2: slightly kludgey temporary way that may hurt efficiency a bit; this is
                      3: another "get it out the door for 4.4" release).  The tests at the end of
                      4: the tests file have accordingly been uncommented.  The primary sign of
                      5: the bug was that something like a?b matching ab matched b rather than ab.
                      6: (The bug was essentially specific to this exact situation, else it would
                      7: have shown up earlier.)
                      8: 
                      9: New in alpha3.3:  The definition of word boundaries has been altered
                     10: slightly, to more closely match the usual programming notion that "_"
                     11: is an alphabetic.  Stuff used for pre-ANSI systems is now in a subdir,
                     12: and the makefile no longer alludes to it in mysterious ways.  The
                     13: makefile has generally been cleaned up some.  Fixes have been made
                     14: (again!) so that the regression test will run without -DREDEBUG, at
                     15: the cost of weaker checking.  A workaround for a bug in some folks'
                     16: <assert.h> has been added.  And some more things have been added to
                     17: tests, including a couple right at the end which are commented out
                     18: because the code currently flunks them (complex bug; fix coming).
                     19: Plus the usual minor cleanup.
                     20: 
                     21: New in alpha3.2:  Assorted bits of cleanup and portability improvement
                     22: (the development base is now a BSDI system using GCC instead of an ancient
                     23: Sun system, and the newer compiler exposed some glitches).  Fix for a
                     24: serious bug that affected REs using many [] (including REG_ICASE REs
                     25: because of the way they are implemented), *sometimes*, depending on
                     26: memory-allocation patterns.  The header-file prototypes no longer name
                     27: the parameters, avoiding possible name conflicts.  The possibility that
                     28: some clot has defined CHAR_MIN as (say) `-128' instead of `(-128)' is
                     29: now handled gracefully.  "uchar" is no longer used as an internal type
                     30: name (too many people have the same idea).  Still the same old lousy
                     31: performance, alas.
                     32: 
                     33: New in alpha3.1:  Basically nothing, this release is just a bookkeeping
                     34: convenience.  Stay tuned.
                     35: 
                     36: New in alpha3.0:  Performance is no better, alas, but some fixes have been
                     37: made and some functionality has been added.  (This is basically the "get
                     38: it out the door in time for 4.4" release.)  One bug fix:  regfree() didn't
                     39: free the main internal structure (how embarrassing).  It is now possible
                     40: to put NULs in either the RE or the target string, using (resp.) a new
                     41: REG_PEND flag and the old REG_STARTEND flag.  The REG_NOSPEC flag to
                     42: regcomp() makes all characters ordinary, so you can match a literal
                     43: string easily (this will become more useful when performance improves!).
                     44: There are now primitives to match beginnings and ends of words, although
                     45: the syntax is disgusting and so is the implementation.  The REG_ATOI
                     46: debugging interface has changed a bit.  And there has been considerable
                     47: internal cleanup of various kinds.
                     48: 
                     49: New in alpha2.3:  Split change list out of README, and moved flags notes
                     50: into Makefile.  Macro-ized the name of regex(7) in regex(3), since it has
                     51: to change for 4.4BSD.  Cleanup work in engine.c, and some new regression
                     52: tests to catch tricky cases thereof.
                     53: 
                     54: New in alpha2.2:  Out-of-date manpages updated.  Regerror() acquires two
                     55: small extensions -- REG_ITOA and REG_ATOI -- which avoid debugging kludges
                     56: in my own test program and might be useful to others for similar purposes.
                     57: The regression test will now compile (and run) without REDEBUG.  The
                     58: BRE \$ bug is fixed.  Most uses of "uchar" are gone; it's all chars now.
                     59: Char/uchar parameters are now written int/unsigned, to avoid possible
                     60: portability problems with unpromoted parameters.  Some unsigned casts have
                     61: been introduced to minimize portability problems with shifting into sign
                     62: bits.
                     63: 
                     64: New in alpha2.1:  Lots of little stuff, cleanup and fixes.  The one big
                     65: thing is that regex.h is now generated, using mkh, rather than being
                     66: supplied in the distribution; due to circularities in dependencies,
                     67: you have to build regex.h explicitly by "make h".  The two known bugs
                     68: have been fixed (and the regression test now checks for them), as has a
                     69: problem with assertions not being suppressed in the absence of REDEBUG.
                     70: No performance work yet.
                     71: 
                     72: New in alpha2:  Backslash-anything is an ordinary character, not an
                     73: error (except, of course, for the handful of backslashed metacharacters
                     74: in BREs), which should reduce script breakage.  The regression test
                     75: checks *where* null strings are supposed to match, and has generally
                     76: been tightened up somewhat.  Small bug fixes in parameter passing (not
                     77: harmful, but technically errors) and some other areas.  Debugging
                     78: invoked by defining REDEBUG rather than not defining NDEBUG.
                     79: 
                     80: New in alpha+3:  full prototyping for internal routines, using a little
                     81: helper program, mkh, which extracts prototypes given in stylized comments.
                     82: More minor cleanup.  Buglet fix:  it's CHAR_BIT, not CHAR_BITS.  Simple
                     83: pre-screening of input when a literal string is known to be part of the
                     84: RE; this does wonders for performance.
                     85: 
                     86: New in alpha+2:  minor bits of cleanup.  Notably, the number "32" for the
                     87: word width isn't hardwired into regexec.c any more, the public header
                     88: file prototypes the functions if __STDC__ is defined, and some small typos
                     89: in the manpages have been fixed.
                     90: 
                     91: New in alpha+1:  improvements to the manual pages, and an important
                     92: extension, the REG_STARTEND option to regexec().