» Refer to the installation instructions when building from source.
» Be sure to read the user manual for this version.
New or improved syntax files for the following languages:
New translations
UTF-16 support
JOE can now edit UTF-16BE and UTF-16LE files. It does this by converting them to UTF-8 during load and back to UTF-16 during save.
Within JOE, native byte order is called UTF-16 and reversed order is called UTF-16R.
If you change the encoding (with ^T E) between UTF-8, UTF-16 and UTF-16R, JOE will convert the file to the desired encoding on save.
New regular expression engine
Old one was a recursive matcher, new one is compiled Thompson NFA matcher
JOE now supports full regular expressions (but as before special characters are escaped by default), so:
JOE also supports the standard regular expression syntax where these characters are not escaped.
Standard syntax regular expressions are now used in the ftyperc
file (which is used to determine the syntax of the file by
inspecting its contents)
Submatches within regular expressions can now be any size (up to the size of the disk!). Before this, they were limited to 16K.
Case conversion allowed in replacement string, as in sed: everything between \U and \E converted to uppercase.
Unicode improvements
Character class database has been updated to the latest version (Unicode 8.0.0)
Switched to new character class data structure for faster Unicode (uses radix search instead of binary search).
Key sequences in the joerc file are now UTF-8 coded Unicode.
Also you can specify Unicode in hexadecimal like this: U+F123
Note that even if you are using an 8-bit locale, keys are
tranlated to UTF-8 before keymap lookup. This means you
must use the Unicode code for your character in the joerc
file, not the 8-bit code for the character.
Jump to matching delimiter (Ctrl-G) now supports Unicode for word delimiters (for example, within XML tags).
Identifiers within JOE now allow Unicode. For example, variables at the math prompt and JOE macros can use any letter.
JOE now displays Unicode combining characters properly
Syntax files are now UTF-8 coded and support Unicode syntax.
Character lists in syntax files and search strings (regular expressions) now provide access to the Unicode category database and provide some other useful character classes:
Use \x{f123} to specify a particular character.
Use \p{Lu} to specify a Unicode character class: any one of L, Lu, Ll, Lt, Lm, Lo M, Mn, Mc, Me N, Nd, Nl, No P, Pc, Pd, Ps, Pe, Pi, Pf, Po S, Sm, Sc, Sk, So Z, Zs, Zl, Zp C, Cc, Cf, Cs, Co, Cn See: ftp://ftp.unicode.org/Public/5.1.0/ucd/UCD.html#General_Category_Values
Use \p{Cherokee} to specify any character from a named Unicode block.
\d for a digit, \D for not a digit
\w for a word character, \W for not a word character
\s for a space character, \S for not a space character
\i for identifier start character, \I for not identifier start character
\c for identifier continuation character, \C for not identifier continuation character
Code clean up
Switch to ptrdiff_t for memory offsets and off_t for file offsets (prior to this, int and long were used). Now you can edit files larger than 4 GB on 32-bit systems.
Give up trying force all strings to "unsigned char *" so that the code is less weird.
Clean up code so that we get a clean compile even with many more warnings enabled. Going forward this helps find real bugs by highlighting new warnings.
Remove C usage which is illegal in C++ so that JOE can be compiled by C++ compilers as well as C compilers. This is useful because C++ compilers sometimes warn about issues that C compilers miss.
Bugs fixed
Fix bug where \ was not parsed correctly within syntax file character lists unless it was at the end of the string.
Fix bug where position cursor history operations would mix pointers between different buffers if user had switched buffers in a window.
Fix bug where lockup would happen if you try querysave when the only buffer left is the startup log.
Default locale
Improve performance where JOE would seem to lock up if you tried to reformat a very long single word due to O(n^3) algorithm.
Prevent filt and blkdel from modifying read-only buffers. This could happen if you run them from modifyable buffer but with block set in a read-only buffer.
Fixed issue when recording paste in a macro. If you tried to play the macro, the pasted text is not inserted and JOE is stuck waiting for the bracketed paste end string.
Fixed issue where syntax could not be set on command line with -syntax.
Minor enhancements
Tab completion now works for the command and its file arguments after '!' in file prompts. Tab completion now works for the filename after '>>' in file write prompts.
Tab completion now handles directory and file names with spaces in them.
Backspace now jumps back to parent menu in ^T submenus (and remembers the cursor position within the parent)
Macros after :def are now allowed to cross lines in the joerc file
Make ^K ^SPACE same as ^K SPACE
Quoted insert of TAB always inserts a TAB character, even when smart indent is enabled.
Add options to control sending of bracketed paste mode command to terminal emulator (brpaste) and detection of paste by timing (pastehack).
Modified ftyperc file syntax to reduce redundancy
Added file type (as defined in ftyperc) setting option. For example, with "joe -type c fred" JOE will assume fred is a C language file. Use ^T F to change the file type from within JOE.
Highighter enhancement: when % is used in place of a character list, it matches the save_c delimiting character as-is (vs. & which matches the opposite character). For example, if save_c has {, then % matches { while & matches }. This allows JOE to highlight q{hello { there } } in Perl.
jmacs fixes:
^X b / ^X ^B were reversed
^X 0 printed an exit message for no reason
^X 0 now can pop shell windows
M-^ deleted indentation but did not join with previous line
Ignore case for letter commands: ^X i and ^X I are the same
Fix bug where regular expressions were not working in incremental search when wrap is enabled (which is the case in jmacs).
ESC g (grep/find) and ESC c (compile) improvements
Tab completion now works for the command and its arguments
Change to the current directory before running the command
Show the current directory in the compile window
Show the exit status in the compile window
Provide more consistent window setup during compile
Parse "Entering directory `/home/xxxxxx'" messages to determine the directory containing the file with an error message.