Changelog:
changes from version ?? to 1.1:
* user interface add key = to change the registers
A,B,C,D,E,H,L,IX,IY,A',B',C',D',E',H',L',I,R
* display and update the cycle counter and the refresh register
* BUG in disasm.c f_ld LD rr,(nn) [only 1 byte loaded] fixed!
* BUG in disasm.c f_ex EX (SP),IX break missing ==> runs into EX (SP),IY; fixed!
* EI replaced by IFF1 & IFF2 (correct LD A,R & LD A,I)
* user interface @(CPU reset), #(clear memory), *(clear all cpu register),
$(clear tick counter), &(SP init)
* console.c: BUG in c_init() usuage of undefined variable t fixed!
changes from version 1.1 to 1.9:
* IN, OUT, INI, OUTD, ... all implemented
* console.c: BUG(?) in c_init() if ICANON cleared MIN should bet set to 1
* BUG in disasm.c f_add all 16-bit addition with carry bit :( fixed!
* declaration of an assembler label may be appended by a colon
* if PC, SP, and MEMP input is requested current values are displayed in hexa
* HALT really halts the CPU now if interrupt is disabled else set runmode false
* user interface !(NMI) implemented
* user interface . puts compiled instructions into memory and increases addr
* user interface = = to change 16-bit registers BC,DE,HL,BC',DE',HL',IX,IY
* BUG in disasm.c f_adc wrong carry/h/v flag calculation fixed!
* BUG in disasm.c f_sbc wrong carry/h/v flag calculation fixed!
* interrupt behaviour simulated, except IM=0 (probably tick count wrong)
* user interface D(dump/disassemble into file) toggle DI/EI to key ^ changed
* PC and SP defaults into hexa, CALL, JP and RST in disassembling format, too
* user interface " (protocol instructions executed)
* bank switching logic implemented (still experimental)
* BUG in disasm.c f_ld if 16 bit load from 0xffff adress OVERFLOW fixed!
* disasm.c: f_sll opcode CB30-CB37 implemented
* warnings and error_msg now occupy only the very last line (row 23)
* user interface % sets a breakpoint with a run counter at current PC
* breakpoints (address & counter) are displayed in row 22
* pseudo instruction EQU in assembler (asm.c) implemented
* CPU Halt-pin/status is displayed in upper right corner if active
* turbo mode improved a lot: 4 levels now (16 min, 380 sec, 38 sec, 2.5 sec)
* disasm.c: f_nop2 all undocumented ED-opcodes implemented
* all the man pages updated
* turbo mode replaced by controling the clock speed/frequency
* the ports are implemented as a map onto one binary file named .Z80_ports
* hardware_clock.c and lcd_display.c provided as examples for port usage
changes from version 1.9 to 2.0beta:
* BUG in inst-dist.c ED63 ld hl,(nn) A_NUM -> A_PODDLE_NUM fixed!
* z80-asm permits in the case of exactly 1 section and inputfilename ends with
.asm the outputfile to be optional (.asm is exchanged to .z80 for outputfile)
* BUG in z80-asm: if start > highest byte in memory undefined outputfile length
* HALT halts the CPU now if interrupt is enabled else set runmode false
necessary for emulator mode ==> user interface & sets a DI on stack
* user interface ! toggle keyboard read disabled/enabled in run mode
user interface ESC now simulates a NMI
* z80-mon has option -E for emulator mode (immediately CPU start)
* SIGTERM, SIGHUP terminates z80-mon, SIGUSR1 causes reset, SIGUSR2 causes NMI,
SIGABORT causes cpu_wait=1 and SIGTRAP makes a CPU dump
* keyboard mapping implemented (used for IN from port stdin)
* simul.c: decode() now handles stray FD and DD prefixes
* all FD and DD are simulated by decode() ==> IXh,IXl,IYh,IYl introduced in
regs.c and regs_token and set N_REGISTERS to 21 , global _ushort ddfdcb_reg
* decoder and executer [decode() and execute.c] totally checked. Now memory
read access separated from memory write access. Usage of DATA pins!
* in disasm.c: all 11 CB-executions and all 8 arith/logic-A-executions bundled
* BUG in disasm.c: f_cp/f_cpd/f_cpi/f_cpdr/... h-flag wrong calculated fixed!
* BUG in disasm.c: f_dec p-flag wrong calculated fixed!
* the f_... functions in execute.c now return a pointer to char, doesn't print
* total memory access encapsulated in memory.c / io-data access simulated
* Makefile updated (*.h) goes now into hardware, make install added
* doc/man pages updated
* in asm.c: write_to_memory(), set_start_address(), get_current_address()
introduced for usage of compile() [now returns the number of bytes writen]
changes from version 2.0b to 2.0c:
* BUG in asm.c: lexical_analysis: multiple label definition possible fixed!
* BUG in compile.c: c_add: ADD IY,IX valid(typo), ADD IY,IY invalid fixed!
* BUG in execute.c: f_alu: ADD,ADC,SBC use second operand (bad nemonic) fixed!
* BUG in execute.c: f_neg: carry-flag always effected (carry=!!A) fixed!
* BUG in execute.c: f_bit: sign-flag effected only if bit 7 tested fixed!
* BUG in execute.c: f_daa: total nonsense! (>=4 independent errors) fixed!
* z80-asm.c Option -l added ==> in asm.c LISTING in pass 2 evaluated
* in execute.c: in disassembling mode jr and djnz instruction offset with sign
* user interface t enables base switch for disassembling constants
* user interface L now can read true binary files into memory
* user interface a,b changed to u,v and e deleted [reserved for register]
* pseudo instructions DEFM, ALIGN in assembler (asm.c) implemented (sll,too)
* Z80-Asm accepts integers to base 16 also if they have a h or H postfix
* hash table structur/allocation optimized and option -c in z80-asm.c added
changes from version 2.0c to 2.0:
* pseudo instruction END added and max label length to be 63 documented
* in asm.h: better interface (highest_address,generated_bytes,set_compile_pass)
* in asm.c: compile(void) now detects an EOF (take_line reads exactly 1 line)
changes from version 2.0 to 2.1
* in asm.c: compile(char *txt) ==> take_line(..) not longer obligate!
* ticks encapsulated ==> wait_tics(AMOUNT) and quartz.c/quartz.h introduced
* -DEFINES in Makefile and hardware/z80-ctc.c (experimental/beta Z80_CTC)
* BUG in compile.c: IX-/ IY-displacements must be < 128 and >= -128 fixed!
* in asm.c: in compile() all error_messages bundled (do we like to translate?)
* daisy_chain realized / more realistic interrupt handling by hardware
* bank switching logic re-implemented (still all banks must have same size)
* in regs.c register names to upper case (better to distinguish from addresses)
* user interface j switches relative versa absolute address (disasm jr/djnz)
* several small errors/bugs fixed
* user interface j enables now labeled address disassembling for jp/call/jr
* user interface j enables relative disassembling via two-pass disassembling
* z80-mon with -E now reads CPU status from .CPU and finishes if HALT in DI
* algebraic expressions evaluation in asm.c: valid operators ~,&,|,^,+,-,*,/,%
* in asm.c: @ is a special 'label' indicating the current PC
* in asm.c: explicit binary constants must be lead-in by # replacing % prefix
additionaly (to the 0x) the $ prefix for hexadecimal numbers is availible
* BUG in compile.c: c_ret: typo in case PO wrong machine code generated fixed!
* several DJGCC compilation bugs removed (missing includes,defines,typos)
changes from version 2.1c to 2.2:
* BUG in asm.c: convert_arg: memmove deletes IX or IY but needed later fixed!
* BUG in execute.c f_out/f_in: ts[] resp. tf[] must be at least 5 chars fixed!
* BUG in compile.c c_ld: LD HL,(addr) should generate opcode 2A not ED6B fixed!
* BUG in compile.c convert_arg: char consts must not be converted to upper case
fixed!
* BUG in compile.c convert_arg: missing {} for for-loop body which results in
all strings became nilstrings fixed!
* BUG in compile.c convert_arg: bug in string parsing, wrong indices fixed!
* BUG in execute.c f_ld: LD (addr),rr puts only lower byte into memory fixed!
* BUG in asm.c: convert_arg: (IX+-num) num argument was not expected to be an
expression. test_number() <-> parse_expr() fixed!
* BUG in asm.c: convert_arg: arg->label undefined! Default to false now fixed!
* BUG in asm.c: lexical_analysis: overflow possible for 8-bit arguments in 1th
pass if labels are used in expressions fixed!
* BUG in asm.c: lexical_analysis: pseudo instructions I_EQU, I_DEFM, I_ORG,
I_ALIGN also needs the current address as
argument in pass 1 if it is a label fixed!
* BUG in asm.c: lexical_analysis: parsing inside '..' was finished by delimiter
(now new error message UNS introduced) fixed!
* BUG in z80-mon.c: main: init_ctc() must be before reset_cpu() because
set_cpu_pin() may call set_ctc_pin()! fixed!
* BUG in execute.c: ix_iy_disp: _ushort a must be casted to signed char fixed!
* BUG in execute.c: reg_str: in case A_PODLE_IX_PLUS, A_PODLE_IY_PLUS argument
arg must be casted to signed char fixed!
* in asm.c: if more than 4 bytes opcodes (e.g. DEFS) only the first 4 are shown
in the assembler listing
* in asm.c: new error message: Invalid character argument for invalid character
tokens containing more than 1 (or none) character
* in all places/files exchanged DEFM <--> DEFS (defs <--> defm). Historical
DEFM is used to put messages (strings) in memory and
DEFS is used to reserve storage (in memory) in Zilog Z80 assembler language
changes after version 2.2:
* BUG in instr.c: DEFS and DEFM must stay ordered lexicographicly fixed!
* BUG in asm.c: compile: in printf if i>4 maximum must be 4 not 12 fixed!
* BUG in asm.c: compile: in I_ORG last must be set to address fixed!
* in memory.c: set_default_byte() introduced to set the default memory value
* in z80-asm new option -f XX (this option defines DEFS & ORG gaps to be XX)
* BUG in asm.c: logic when to calculate EQU is wrong!! we need to store the
expression in the hash table NOT FIXED! "(pruchod != 11)"
* BUG in asm.c: the ALIGN statement must produce fill bytes, too (-f XX) fixed!
* BUG in asm.c: overwriting code warning must depend on fill byte fixed!
* BUG in asm.c: test_number: if base is known (prefix) number may be start
with non-digit character fixed!
* BUG in asm.c: compile I_ALIGN address is already increased by out() fixed!
* BUG in z80-asm.c: init: now it has parameter fill_byte to init memory fixed!
* in asm.c/expression.c/hash.c: new evaluation logic for EQU and expressions
implemented. Their values are now evaluated in pass 2 recursivly
* in z80-mon.c: ask() and ask_x() accept now all number representations like
z80-asm (even the '?') except that any sign is forbidden
* pseudo instructions DEFL in assembler (asm.c) implemented
* BUG in z80-asm.c: parse_arg: ignores filename parameter if one can be
constructed from source filename fixed!
* BUG in compile.c: out() should be performed to have correct PC if signed or
unsigned overflow of 8-bit argument occurs.(ret=8,9) fixed!
* in expression.c: new operators !, <<, >>, ==, !=, >, <, >=, <=, &&, ||
* BUG in hash.c: last_label_reusable: uniq must be negated fixed!
* in asm.c: DEFL lables now multiple in hashtable (distinguished by lineno)
* in asm.c: COND/ENDC enables conditionel compiling
* BUG in execute.c: f_in_block, f_out_block: F_sign, F_Z not effected and
if B!=0 F_Z not cleared fixed!
* BUG in Makefile: now manuals are in doc/man not in doc fixed!
* BUG in asm.c: convert_arg: indirect addressing was wrongly recognized fixed!
* BUG in hash.c: is_in_table: for non-uniq lables (DEFL) wrong element fixed!
* BUG in hash.c: add_to_table: ->uniq undefined fixed!
* in expression.c: new operators ** , // , ? and right operand of << and >>
is now interpreted as unsigned
* BUG in expression,c: parse_expr(): all expressions must be tried to be
evaluated in PASS 1 (needed for ORG, DEFS, CONDC) fixed!
* in ports.c: cpu_pin iorq used, and the variables ADDRESS and DATA are set.
* in hardware/README: explanation how to implement further add-ons
* z80-mon.c: main: cpu_pins busrq and busack are now served
* z80-cpu.c/ports,c: Z80-CTC port access now wired at Z80-CPU
* BUG in ports.c: in_byte(): missing DATA bus assign from *data fixed!
changes after version 2.3:
* BUG in Makefile: DEFINES are not put to in $(MAKE) hardware fixed!
* hardware: logic_analzyer implemented (to view the Z80 pins)
* in decode.c: refresh_cycle() mreq- and rfsh-pin used and I<<8|R onto ADDRESS
* in z80-cpu.c: machine cycle counter (triggered by wait_tics)
* BUG in asm.c: no_para[]: RST needs 1 Parameter, not 0 fixed!
* BUG in execute.c: cp_block: 3 tics missing fixed!
* in z80-mon.c: machine cycle counter displayed
* in z80-cpu.c: busrq is now acknowledged at beginning of each machine cycle
* with dummy.c print_ticks() reintroduced to mask it in non z80-mon programs
using acknowledge_bus_request()
* BUG in hardware/z80-ctc.c: read_word(): fp may be 0 fixed!!
* only if !cpu_is_in_disassemble then set_cpu_pin() , acknowledge_bus_request()
and ADDRESS are used (databus can be used if cpu_is_in_disassemble mode)
* port_buffer for OLD_STYLE port_access via Z80_PORTS put in hardware now
* z80-global im Makefile als dependence of z80-cpu.h
* SSL_ZDS trigger in hardware
* BUG in execute.c: f_inc/f_dec: 16 bit registers shouldn't change flags fixed!
* BUG in execute.c: f_rrd: lower nibble in accu set wrongly fixed!
* BUG in execute.c: f_adc16/f_sbc16: zero_flag wrongly calculated fixed!
* BUG in execute.c: f_inc/f_dec: PV_flag wrongly calculated fixed!
* in z80-asm.c und z80-mon.c option parsing simplified und equaly processed.
* BUG in execute.c: add,adc,sub,sbc,cp,f_adc16,f_sbc16: underflow check for
< -127 , < -32767 missing (V-Flag wrongly computed) fixed!
* BUG ins asm.c: convert_arg(): arg->text isn't initialized to 0. fixed!
* BUG in regs.h: missing #include "z80-global" to define seznam_type fixed!
* BUG in execute.c: f_neg: PV_flag wrongly calculated fixed!
* BUG in execute.c: f_neg: H_flag setiing typo: F_N must be F_NH fixed!
* BUG in execute.c: f_inc/f_dec: PV_flag wrongly calculated for 8-bit register
(last fix affected only 8-bit memory access) fixed!
* BUG in execute.c: add/adc/sub/sbc/cp/adc16/sbc16: V_flag inverted mapped on
P_flag fixed!
* in all Makefile: -malign-function=0 replaced by -falign-function=0 and
Makefile-Variable CC set to gcc.
* doc/execute_timing.tex to doc/execute_timing.txt renamed
* BUG in z80-mon.c: main: counter b & s for argv wrongly increased fixed!
* BUG in memory.c: init_banks: map must be 0-pointer for first realloc fixed!
* BUG in memory.c: init_banks: tok[3] must be tok[4] in two lines fixed!
* BUG in memory.c: init_banks: superfluious else prohibited bank[id].back_up
allocation. fixed!
* BUG in memory.c: init_banks: strtoul(..) needs check of *p if p is defined
after scanning tok2 and tok4 for numbers fixed!
* BUG in memory.c: _uchar memory_at: logic for bank_access is missing fixed!
* in hardware/banks-*: new bank-mask notation introduced
* in doc/man/z80-banks.5 default bank after reset/initialisation documentated.
* in memory.c bank-mask must now be a 2 hexadecimal number
* in doc/man/z80-mon.1 execution-mode preciser defined.
* in z80-cpu.c/z80-cpu.h: new cpu-flag cpu_is_in_x_mode defined
* in z80-mon.c: main: cpu_is_in_x_mode is set if 'Execute instruction'-Key
* in execute.c: cpu_is_in_x_mode used for jr/djnz/halt/cp-, ld-, out-, in-
block instructions used
This release addresses two security issues:
CVE-2014-3580: mod_dav_svn DoS from invalid REPORT requests.
CVE-2014-8108: mod_dav_svn DoS from use of invalid transaction names.
Version 1.8.11
(15 Dec 2014, from /branches/1.8.x)
http://svn.apache.org/repos/asf/subversion/tags/1.8.10
User-visible changes:
- Client-side bugfixes:
* checkout/update: fix file externals failing to follow history and
subsequently silently failing (issue #4185)
* patch: don't skip targets in valid --git difs (r1592014, r1592034)
* diff: make property output in diffs stable (r1589360)
* diff: fix diff of local copied directory with props (r1619380, r1619393)
* diff: fix changelist filter for repos-WC and WC-WC (r1621978, r1621981)
* remove broken conflict resolver menu options that always error out
(r1620332)
* improve gpg-agent support (r1600331, r1600348, 1600368, r1600563,
r1600781)
* fix crash in eclipse IDE with GNOME Keyring (issue #3498)
* fix externals shadowing a versioned directory (issue #4085)
* fix problems working on unix file systems that don't support
permissions (r1612225)
* upgrade: keep external registrations (issue #4519)
* cleanup: iprove performance of recorded timestamp fixups (r1633126)
* translation updates for German
- Server-side bugfixes:
* disable revprop caching feature due to cache invalidation problems
(r1543594, r1619774, r1619105, r1619118, r1619153, r1619802)
* skip generating uniquifiers if rep-sharing is not supported (r1561426)
* mod_dav_svn: reject requests with missing repository paths (r1643409)
* mod_dav_svn: reject requests with invalid virtual transaction names
(r1643437)
* mod_dav_svn: avoid unneeded memory growth in resource walking
(issue #4531)
Developer-visible changes:
- General:
* make sure all members of the repos layer notify struct are valid,
fixes crashes in API users using all members (r1616131)
* properly generate a version resource when building on Windows (r1542610,
r1564576, r1568180)
* fix LIBTOOL_M4 and LIBTOOL_CONFIG variable not be evaluated properly
during a unix build (r1637826)
* allow the use of libtool 2.4.3 (r1640862, r1640873, r1643793)
This is an unscheduled bugfix release containing two security fixes
for issues we uncovered in both Git and Mercurial for CVE-2014-9390.
Users on Mac and Windows are encouraged to upgrade.
context: stop setting None for modified or added nodes
darwin: omit ignorable codepoints when normcase()ing a file path
encoding: add hfsignoreclean to clean out HFS-ignored characters
largefiles: don't actually remove largefiles in an addremove dry run
log: fix log -f slow path to actually follow history
log: fix log revset instability
manifest: disallow setting the node id of an entry to None
pathauditor: check for Windows shortname aliases
pathauditor: check for codepoints ignored on OS X
rebase: ignore negative state when updating back to original wc parent
update: add tests for untracked local file
update: don't overwrite untracked ignored files on update
* We used to allow committing a path ".Git/config" with Git that is
running on a case sensitive filesystem, but an attempt to check out
such a path with Git that runs on a case insensitive filesystem
would have clobbered ".git/config", which is definitely not what
the user would have expected. Git now prevents you from tracking
a path with ".Git" (in any case combination) as a path component.
* On Windows, certain path components that are different from ".git"
are mapped to ".git", e.g. "git~1/config" is treated as if it were
".git/config". HFS+ has a similar issue, where certain unicode
codepoints are ignored, e.g. ".g\u200cit/config" is treated as if
it were ".git/config". Pathnames with these potential issues are
rejected on the affected systems. Git on systems that are not
affected by this issue (e.g. Linux) can also be configured to
reject them to ensure cross platform interoperability of the hosted
projects.
* "git fsck" notices a tree object that records such a path that can
be confused with ".git", and with receive.fsckObjects configuration
set to true, an attempt to "git push" such a tree object will be
rejected. Such a path may not be a problem on some filesystems
but in order to protect those on HFS+ and on case insensitive
filesystems, this check is enabled on all platforms.
A big "thanks!" for bringing this issue to us goes to our friends in
the Mercurial land, namely, Matt Mackall and Augie Fackler.
Also contains typofixes, documentation updates and trivial code clean-ups.
Changes since v2.2.0 are as follows:
Hartmut Henkel (1):
l10n: de.po: fix typos
Jeff King (8):
unpack-trees: propagate errors adding entries to the index
read-tree: add tests for confusing paths like ".." and ".git"
verify_dotfile(): reject .git case-insensitively
t1450: refactor ".", "..", and ".git" fsck tests
fsck: notice .git case-insensitively
utf8: add is_hfs_dotgit() helper
read-cache: optionally disallow HFS+ .git variants
fsck: complain about HFS+ ".git" aliases in trees
Johannes Schindelin (3):
path: add is_ntfs_dotgit() helper
read-cache: optionally disallow NTFS .git variants
fsck: complain about NTFS ".git" aliases in trees
- Add two patches, to avoid
error: incompatible types when assigning to type '__off_t' from type 'fpos_t'
fp_->_offset = u.f;
- Remove one patch patch-af, which had a comment, not necessary for now
" also look in *.mk files as makefiles. bump pkg revision."
(upstream)
- Update 4.2 to 4.6
----------------------
* Noteworthy changes in release 4.6 (2012-02-03) [stable]
** Bug fixes
- lid -L no longer mishandles open-ended ranges like "..2" and "2.."
- lid's -d, -o and -x options now work properly
* Noteworthy changes in release 4.5 (2010-06-17) [stable]
* Noteworthy changes in release 4.4 (2009-10-30) [beta]
------------
** New features
mkid and xtokid accept a new option --files0-from=FILE, to make them
process only the files named in FILE. FILE must contain a list of
NUL-terminated file names.
** Bug fixes
mkid and xtokid now accept language specific options via the command line.
* Noteworthy changes in release 4.3.92 (2008-10-18) [beta]
--------------
** Bug fixes
fid: avoid a buffer overrun
handle failed allocation, e.g., by strdup
avoid potential realloc overflow
mkid: avoid an infloop on some .el files
all programs: detect write error on stdout
** Miscellaneous improvements
revamp code and infrastructure, bringing it closer to coreutils' standards
add automatically-generated man pages
mkid recognize more suffixes: .ac, .mk, .bz2, .lzma.
idutils is now licensed under the General Public License version 3
or later (GPLv3+).
- Remove patch-aa, no file found, no similar lines found.
- Add LICENSE= public-domain, following line is found in README
Berkeley Yacc is in the public domain.
(upstream)
- Update from 20050813 to 20141128
- CHANGLOG found but too huge to quote here.
ccache 3.2.1
------------
Release date: 2014-12-10
Bug fixes
~~~~~~~~~
- Fixed regression in temporary file handling, which lead to incorrect
permissions for stats, manifest and ccache.conf files in the cache.
- `CACHEDIR.TAG` files are now created in the [0-9a-f] subdirectories so that
ccache.conf is not lost in backups.
- Made the default cache size suffix `G`, as previously documented.
- `-fdiagnostics-color=auto` is now passed to the compiler even if stderr is
redirected. This fixes a problem when, for instance, a configure test probes
if the compiler (wrapped via ccache) supports `-fdiagnostics-color=auto`.
- Added missing documentation for `max_files` and `max_size` configuration
options.
-----
8.0.1
-----
* Issue #296: Restore support for iteration over parse_version result, but
deprecated that usage with a warning. Fixes failure with buildout.
---
8.0
---
* Implement `PEP 440 <http://legacy.python.org/dev/peps/pep-0440/>`_ within
pkg_resources and setuptools. This change
deprecates some version numbers such that they will no longer be installable
without using the ``===`` escape hatch. See `the changes to test_resources
<dcd552da64 (chg-setuptools)/tests/test_resources.py>`_
for specific examples of version numbers and specifiers that are no longer
supported. Setuptools now "vendors" the `packaging
<https://github.com/pypa/packaging>`_ library.
* Changes in the m17n library 1.7.0
** The mechanism of fallback input method is implemented. For that, a
new variable fallback-input-method and new commands switch-im,
push-im, pop-im are supported.
1.5.0
~~~~~
Improvements
------------
* When an import error happens ``testtools.run`` will now show the full
error rather than just the name of the module that failed to import.
(Robert Collins)
1.4.0
~~~~~
Changes
-------
* ``testtools.TestCase`` now inherits from unittest2.TestCase, which
provides a ``setUpClass`` for upcalls on Python 2.6.
(Robert Collins, #1393283)
1.3.0
~~~~~
Changes
-------
* Fixed our setup.py to use setup_requires to ensure the import dependencies
for testtools are present before setup.py runs (as setup.py imports testtools
to read out the version number). (Robert Collins)
* Support setUpClass skipping with self.skipException. Previously this worked
with unittest from 2.7 and above but was not supported by testtools - it was
a happy accident. Since we now hard depend on unittest2, we need to invert
our exception lookup priorities to support it. Regular skips done through
raise self.skipException will continue to work, since they were always caught
in our code - its because the suite type being used to implement setUpClass
has changed that an issue occured.
(Robert Collins, #1393068)
1.2.1
~~~~~
Changes
-------
* Correctly express our unittest2 dependency: we don't work with old releases.
(Robert Collins)
1.2.0
~~~~~
Changes
-------
* Depends on unittest2 for discovery functionality and the ``TestProgram`` base
class. This brings in many fixes made to discovery where previously we were
only using the discovery package or the version in the release of Python
that the test execution was occuring on. (Robert Collins, #1271133)
* Fixed unit tests which were failing under pypy due to a change in the way
pypy formats tracebacks. (Thomi Richards)
* Fixed the testtools test suite to run correctly when run via ``unit2``
or ``testtools.run discover``.
* Make `testtools.content.text_content` error if anything other than text
is given as content. (Thomi Richards)
* We now publish wheels of testtools. (Robert Collins, #issue84)
1.1.0
~~~~~
Improvements
------------
* Exceptions in a ``fixture.getDetails`` method will no longer mask errors
raised from the same fixture's ``setUp`` method.
(Robert Collins, #1368440)
CHANGELOG
2014/11/05 - 0.8
* Issue #22457: Honour load_tests in the start_dir of discovery.
2014/10/31 - 0.7.1
Fix for 0.7.0 being broken. Also switches from both manual and entrypoint
scripts to just entrypoint scripts to reduce the possibility of similar
oversights in future.
2014/10/31 - 0.7.0
This release contains backports from cPython 3.5 of all (I think) commits since
2010, as of today.
2014/10/28 - 0.6.0
Many thanks to Mark Roddy and Ezio Melotti who contributed substantially to
this release.
* Changed supported Python versions to start at 2.6, and include all released 3.x and pypy. (Robert Collins)
* Invoking unit2 without args starts test discovery
* Added TestCase.assertWarns and TestCase.assertWarnsRegexp context managers
* Fix Python issue 9926. TestSuite subclasses that override __call__ are called correctly.
* Removed unused maxDiff parameter from TestCase.assertSequenceEqual.
* DeprecationWarning for unsupported result objects (missing addSkip method) became RuntimeWarning.
* Addition of TestCase.assertWarns as a context manager.
extras is a set of extensions to the Python standard library, originally
written to make the code within testtools cleaner, but now split out for
general use outside of a testing context.
CHANGES:
## Version 1.2.3.0
* Silence integer literal overflow warning
* Add support for GHC 7.10 `integer-gmp2` & `Natural`
* Add instance for Data.Void
* Make the SSE .cabal flags manual
* Add an upper bound on bytestring
changelog:
1.2.0.2
* Bumped lower bound on deepseq to 1.4 for compatibility with the
upcoming GHC 7.10
1.2.0.1
* Fixed a buffer overflow in rendering of large Integers
(https://github.com/bos/text/issues/99)
Version 0.8.4 (2014-12-11)
- Fix issue related to using argcomplete in a REPL environement. Thanks
to @wapiflapi (pull request #91).
Version 0.8.3 (2014-11-09)
- Fix multiple issues related to using argcomplete in a REPL environement.
Thanks to @wapiflapi (pull request #90).
Version 0.8.2 (2014-11-03)
- Don't strip colon prefix in completion results if COMP_WORDBREAKS does
not contain a colon. Thanks to @berezv (pull request #88).
1.13 rurban (2013-04-06)
- removed diag before each big t/11keywords.t loop
- added suggested keywords from RT #62382:
*_ and DATA to @FileHandles (the deprecated lowercase: stdin, stdout, stderr not),
import unimport to @Functions,
BINMODE CLEAR CLEARERR CLONE CLONE_SKIP CLOSE DELETE EOF
ERROR EXISTS EXTEND FDOPEN FETCH FETCHSIZE FILENO FILL FIRSTKEY FLUSH
GETC NEXTKEY OPEN POP POPPED PRINT PRINTF PUSH PUSHED READ READLINE
SCALAR SEEK SETLINEBUF SHIFT SPLICE STORE STORESIZE SYSOPEN TELL
TIEARRAY TIEHANDLE TIEHASH TIESCALAR UNREAD UNSHIFT UNTIE UTF8 WRITE to @TieIOMethods,
can isa DOES VERSION to @UNIVERSALMethods,
@EXPORT @EXPORT_OK @EXPORT_FAIL
@EXPORT_TAGS _push_tags _rebuild_cache as_heavy export export_fail
export_fail_in export_ok_tags export_tags export_to_level heavy_export
heavy_export_ok_tags heavy_export_tags heavy_export_to_level
heavy_require_version require_version to @ExporterSymbols (Zsb n Ambrus)
- Moved exp from @Barewords to @Functions
- Added $^CHILD_ERROR_NATIVE $^GLOBAL_PHASE $^LAST_FH $^MATCH $^PREMATCH
$^POSTMATCH $^UTF8CACHE $^WIN32_SLOPPY_STAT to @Scalars
- Added English names for %!, @F (perlrun) and @ARG for @_ (Zsb n Ambrus)
- Added %+ %- (Zsb n Ambrus) and $LAST_SUBMATCH_RESULT
0.12 2013-05-11
* fix test that breaks in 5.17.4 (RT#80070, Father Chrysotomos)
* Require ExtUtils::Depends 0.302 for win32 support (RT#77311), Alexandr
Ciornii)
0.11 Mon, 18 Jun 2012 17:01:00 +0100
* Document get_lex_stuff and clear_lex_stuff.
* Document return type of C version of set_linestr.
1.2 Fri Apr 11 06:22:38 CEST 2014
- perl5porters broke Async::Interrupt, BDB, EV, IO::AIO, OpenCL
without warning by switching the meaning of USE_SOCKETS_AS_HANDLES
in 5.18. What's so attractive about giving a shit about backwards
compatibility - I will never understand.
- Add Full dependency to devel/p5-Type-Tiny
(Configure warns if not)
(upstream)
Update 2.02 to 2.11
-------------------
Dave Cross <dave@dave.org.uk> 2014-10-01
Finish removing Moose
Dave Cross <dave@dave.org.uk> 2014-10-01
File permission fixes
Dave Cross <dave@dave.org.uk> 2014-10-01
Ported from Moose to Moo
Dave Cross <dave@dave.org.uk> 2014-01-10
Added .travis.yml
---------------------
New in CSSC-1.4.0, 2014-07-11
* New stable release, almost identical to the test release
1.3.2 - the only changes are version numbering and updates
to the ChangeLog and NEWS files. The previous stable
release was 1.3.0, so the following two sections of this
file set out what has changed since 1.3.0.
New in CSSC-1.3.2, 2014-06-30
* This release is based on updated versions of gnulib and of
the googletest unit test framework (the test framework is of
course not linked into the installed CSSC binaries, it is
just used for unit tests).
* This release uses a new tar format. Up to now we have used
the v7 Unix tar format, but this does not cope correctly
with filenames longer than 99 characters, and we now have
one of those. We now use the the ustar format defined by
POSIX 1003.1-1988. If you need to unpack a CSSC
distrubition with a version of tar which only deals with the
v7 format, unpack the distribution somewhere else, rename or
delete the long files (they are not essential to the build
process) and re-pack it. This new tar format will likely
also be used for all future releases.
* Fix Savannah bug 33633, prs -d":GB:" coredumps when
expanding %I%.
New in CSSC-1.3.1, 2011-05-10
* prs now prints ignored deltas when expanding the :DI:
keyword. The expansion of :DI: varies between SCCS
implementations, as described in the manual. This partly
fixes Savannah bug #33154: SCCS versions vary wrt ignored
deltas (that is, this change covers only the prs binary).
* Fix Savannah bug bug #33153: "prs" includes "AUTO NULL
DELTA"s. "prs -r" will no longer print any additional
deltas that happen to share the same timestamp with the last
matching delta.
* If the SID identified by "prs -r" is not found, prs now
exits with status 1, not 0. This fixes SourceForge bug
#1300722.
* The headers produced by sccsdiff have changed. A different
header is used when the -p option is specified and when
it's not. The new behaviour matches OpenSolaris 2009.06.
* The ordering of the ouput of "sccs unedit" is now different;
the new order matches OpenSolaris 2009.06.
* "get -L" will now send the ID of the gotten SID to stderr,
which better matches OpenSolaris 2009.06. The indentation
of the delta summary has also been adjusted.
* To pass an empty list of MRs to admin, you now have to
provide an argument (as opposed to omitting it). For
example, "admin -i/dev/null -fv -m' ' s.foo". This aligns
better with OpenSolaris 2009.06 but it is a
backward-incompatible change.
-----------------------
Sat July 26, 2014: Released Check 0.9.14
based on r1174 (2014-07-03 18:43:49 +0000)
* Changes to (unofficially for now) support Solaris and AIX platforms.
* Fix issue with checked teardown functions in CK_NOFORK mode. If
the teardown function called an assertion that failed, an
infinate loop would result. Bug#98. As a result of the change, for now
checked teardown functions that fail in CK_NOFORK mode will
not cause the test to fail.
* Allow checked fixtures when compiled without fork(). Previously
such setup/teardown functions were disabled, as the full contract
for checked fixtures could not be honored. Checked fixtures
in CK_NOFORK mode are now closer to the contract, and are now
enabled.
* Various code cleanup, including changes for compiling Check on
c++ compilers Clang and g++ (which are less forgiving than their
c counterparts).
* Re-enable internal unit tests, which verify which line numbers are
reported for unit test failures. (The tests were unintentionally
disabled in a previous release).
Fri May 30, 2014: Released Check 0.9.13
based on r1137 (2014-05-26 21:03:09 +0000)
* When a test assertion fails, exit() was invoked. This results in any
registered exit handlers being invoked. A test should not be relied
upon to clean up if it fails. Instead, _exit() will be used to signal
a test failure, to prevent exit handlers from being invoked.
* Fix issue with string formatting in ck_assert_(u)int_* calls, where using
the % operator would be interpreted as a string formatter. Bug #96.
* If the LOG/XML/TAP file name is set to "-" either through the
srunner_set_log() srunner_set_xml() or srunner_set_tap() calls
or through the related environment variables, the logged data will be
printed to stdout instead of a file. This is mainly to support using TAP
for fixtures which expect the results to be reported via stdout. However,
it is available via all of Check's logging modalities.
* For CMake builds, check_stdint.h was not being installed. This is now
fixed.
* Check's example under doc/example now also works for CMake projects
using Visual Studios and MSVC. This includes an example CMake
configuration.
Mon Jan 20, 2014: Released Check 0.9.12
based on r1054 (2014-01-16 23:08:03 +0000)
* Additional unit tests created for internal and external Check
APIs.
* Check now compiled on Windows using MSVC when using CMake and
NMake. All unit tests pass, though shell script based ones need
to be run in the MSYS environment.
* Check now compiles on Windows using Visual Studio 10
when using CMake. check_check passes when run from Visual Studios.
* Always capture the start and end times of tests when using
NO_FORK mode. Previously the end time was not captured,
resulting in arbitrary durations being recorded when tests
failed. Bug #87.
* Added additional configure script checks for support of
timer_create() on the target system. This allows for
OpenBSD to compile and run all Check's unit tests successfully.
* Added a unit test, check_mem_leaks, which can be used against
valgrind to test for memory leaks. No memory leaks were found.
* Added tcase_add_loop_test support in checkm. Patch from patches #46.
* Add support for logging in Test Anything Protocol (TAP) format.
* Refactor Check's assertions to be more like the assert() call in
assert.h, in that static source code analyzers can use gcc attributes
in the header to make assumptions about the flow of the code. See
feature request #29.
* fix ck_assert_ptr_* causing const compilation warnings. Patch from
bug #91.
Wed, Nov 4, 2013: Released Check 0.9.11
based on r856 (2013-11-04 02:09:21 +0000)
* Check's unit tests pass when compiled out of the source tree.
* Check compiles for Windows using the MinGW/msys environment (without using fork), and all unit tests pass.
* Check compiles for Windows using the Cygwin environment, and all unit tests pass.
* Check compiles for Windows using MinGW in Linux (without using fork), and all unit tests pass using wine 1.4.
* Check compiles for Windows using MinGW-w64 in Linux (without using fork), and all unit tests pass using wine 1.4.
* On systems without timer_settimer, use setitimer (if available) to get
subsecond unit test timeouts. If setitimer is unavailable, fallback
on alarm.
