Release 0.6.1 (TL'16) [May 2016]
* Fix "Jump to Source" from PDF context menu
* Fix fine-sync'ing close to PDF line boundaries
* Fix loading of Qt-global translations
* Fix pixel-alignment after zoomFitWindow in continuous mode
* Fix handling of changing scroll bar visibilities in zoomFitWidth and
zoomFitWindow
* Allow horizontal scrolling by Shift+MouseWheel
* Update translations
Release 0.6.0 (TL'16) [April 2016]
* Complete redesign of the PDF previewer using QtPDF by Charlie Sharpsteen
* Fast and responsive preview even at very high magnifications
* New page modes: continuous scrolling and two-column layout
* Text selection/copying from the PDF
* Go back to previous view after clicking on a link using Alt+left arrow
* Fix disappearing spellcheck underline with some fonts/font sizes in Qt 5
* Update the width of the line number display when changing the editor font
* Show a note in the "About Scripts" dialog for disabled script languages
* Explicitly add a "Quit TeXworks" menu item to the global menubar on
Mac OS X
* Make Qt5 the default framework for building TeXworks
* Use Travis CI for continuous integration tests and builds for Windows and
Mac OS X
* Add support for Lua 5.3
Release 0.4.6 (TL'15) [April 2015]
* Implement character-level synchronization by using text searching to
assist SyncTeX
* Lift hardwrap line length limitation and implement (hard) unwrap
* Always show console output tab while typesetting to show progress
* Improve the handling of UTF-8 byte order marks
* Improve the log parsing script
* Improve the handling of script errors
* Work around a race condition that causes lines to become invisible,
subsequently leading to a crash
* Fix incorrectly accumulating indent
* Fix SyncTeX initialization with non-ASCII filenames
* Fix selection of whole lines and when selecting right-to-left
* Fix hangs and crashes when using "Fit to ..." in the PDF view
* Fix a crash when running an invalid system command from scripts
* Fix encoding issues when reading environment variables
* Fix scrolling when searching for multi-line strings
* Fix auto-completion when working with RTL languages
* Improve the handling of Retina displays in Mac OS X
* Fix a crash when opening PDF files from the Windows Explorer
* Provide texworks.appdata.xml for *nix platforms
* Fix compilation with Qt 5
* Improve the CMake build system
* Add support for Lua 5.2
* Move development from Google Code to GitHub
2017.01.10 - GNU nano 2.7.4 "Red dress" undoes deletions in an orderly
manner again (bug was introduced in previous version),
sets the preferred x position for vertical movements
more consistently, avoids some scrolling problems in
softwrap mode, installs the Info manual also when your
system lacks 'makeinfo', and corrects the behavior of
the beginning-of-word anchor (\<) in regex searches.
2016.12.28 - GNU nano 2.7.3 "Ontbijtkoek" wipes away a handful of bugs:
your editor is now able to handle filenames that contain
newlines, avoids a brief flash of color when switching
between buffers that are governed by different syntaxes,
makes the Shift+Ctrl+Arrow keys select text again on a
Linux console, is more resistant against malformations
in the positionlog file, and does not crash when ^C is
typed on systems where it produces the code KEY_CANCEL.
Oh, and it no longer mistakenly warns about editing an
unlocked file just after saving a new one. That's it.
Tastes great with thick butter.
MASTER_SITES= site1 \
site2
style continuation lines to be simple repeated
MASTER_SITES+= site1
MASTER_SITES+= site2
lines. As previewed on tech-pkg. With thanks to rillig for fixing pkglint
accordingly.
Changes in version 1.14:
Version 1.14 is the largest bug hunt ever attempted in GNU ed. Other
goals of version 1.14 are to complete the documentation and to remove
any gratuitous incompatibilities with the POSIX standard. Thanks to Ori
Avtalion for initiating all this with a couple bug reports. ;-)
Byte counts, informative messages, command error messages, and the '?'
and '!' prompts are now written to stdout instead of to stderr. The
standard error (stderr) is now used only for diagnostic messages.
The current address is now correctly set to the addressed line after an
empty insert command.
Fixed inconsistent behavior of the substitute command. It incorrectly
reported 'Invalid pattern delimiter' when the two last delimiters were
omitted after a null regular expression. Now it consistently reports
'Missing pattern delimiter' if the two last delimiters are omitted after
any regular expression (null or not).
's/a/%' has been fixed. It incorrectly replaced 'a' with '%' instead of
using the replacement from the last substitution.
An infinite loop, happening when EOF was found in the middle of a
replacement string, has been fixed.
Ed no longer accepts newlines in the replacement of a 's' command if it
is part of the command list of a global command, because in this case
the meaning of the newline becomes ambiguous. For the same reason, the
last delimiter can't be omitted if the 's' command is not the last
command in the command list.
The substitute command now correctly sets the current address to the
address of the last line on which a substitution occurred, and leaves it
unchanged if no substitution is performed.
A bug in the calculation of address offsets has been fixed. '3 ---- 2'
was calculated as address -2 instead of the correct address 1.
Address ranges with the first address omitted are now accepted.
The current address is now correctly set to the addressed line (or to
the new last line if at EOF) after an empty replacement text in the
change command.
Repeated print suffixes are now rejected. It has been documented that ed
allows any combination of non-repeated print suffixes and combines their
effects.
The substitute command now accepts suffixes in any order.
The 'repeat substitution' command now rejects multiple count suffixes.
The 'p' suffix of the 'repeat substitution' command now toggles all the
print suffixes of the last substitution.
End of file on standard input now behaves as a 'q' command.
The modified status is no longer cleared after writing the buffer to the
standard input of a shell command. (Reported by Jérôme Frgacic).
The descriptions of the 'a', 'c', 'e', 'g', 'i', 'j', 'k', 'm', 'q',
'r', 's', 'u' and 'w' commands in the manual have been fixed.
Most tests in the testsuite have been improved. Bug reporting has been
simplified; only the failed logs and results are kept in the test
directory, which can then be (tarred, compressed, and) attached to the
bug report.
Changes:
8.0.0087 channel callback may get job info after it was deleted
8.0.0088 when a test fails in Setup/Teardown problem is not reported
8.0.0089 various problems with GTK 3.22.2
8.0.0090 cursor moved after last character when using 'breakindent'
8.0.0091 test_help_complete sometimes fails in MS-Windows console
8.0.0092 C indent does not support nested namespaces that C++ 17 has
8.0.0093 not using multiprocess build feature
8.0.0094 if vimrun.exe not found error message is not properly encoded
8.0.0095 problems with GTK 3.22.2 fixed in 3.22.4
8.0.0096 when the input or output is not a tty Vim appears to hang
8.0.0097 if channel callback takes a lot of time Vim is unresponsive
8.0.0098 (after 8.0.0097) can't build on MS-Windows
8.0.0099 popup menu appears above the cursor too often
8.0.0100 file name options may contain non-filename characters
8.0.0101 some options are not strictly checked
8.0.0102 (after 8.0.0101) cannot set 'dictionary' to a path
8.0.0103 may not process channel readahead
8.0.0104 value of 'thesaurus' option not checked properly
8.0.0105 can't tell difference between reading empty line or nothing
8.0.0106 (after 8.0.0100) cannot use a semicolon in 'backupext'
8.0.0107 when reading channel output in timer messages may be missing
8.0.0108 (after 8.0.0107) the channel "drop" option is not tested
8.0.0109 still checking if memcmp() exists, should be everywhere now
8.0.0110 drop command doesn't use existing window
8.0.0111 the :history command is not tested
8.0.0112 tests 92 and 93 are old style
8.0.0113 MS-Windows: dialog for saving changes on the wrong monitor
8.0.0114 coding style not optimal
8.0.0115 when building with Cygwin libwinpthread isn't found
8.0.0116 using CTRl-] in English help language from 'helplang' is used
8.0.0117 parallel make fails
8.0.0118 "make proto" adds extra function prototype
8.0.0119 no test for using CTRL-R on the command line
8.0.0120 channel test is still flaky on OS X
8.0.0121 setting 'cursorline' changes the curswant column
8.0.0122 channel test is still flaky on OS X
8.0.0123 modern Sun compilers define "__sun" instead of "sun"
8.0.0124 internal error for assert_inrange(1, 1)
8.0.0125 not enough testing for entering Ex commands
8.0.0126 display problem with 'foldcolumn' and a wide character
8.0.0127 cancelling completion still formats text
8.0.0128 (after 8.0.0126) display test fails on MS-Windows
8.0.0129 parallel make still doesn't work
8.0.0130 configure uses "ushort" while the Vim code doesn't
8.0.0131 not enough test coverage for syntax commands
8.0.0132 (after 8.0.0131) test fails because of using :finish
8.0.0133 "2;'(" causes ml_get errors in an empty buffer
8.0.0134 null pointer access reported by UBsan
8.0.0135 relative address does not work with closed fold
8.0.0136 when changing indent the wrong fold is opened
8.0.0137 when 'maxfuncdepth' set above 200 nesting is limited to 200
8.0.0138 (after 8.0.0137) small build fails
8.0.0139 (after 8.0.0135) warning for unused argument
8.0.0140 pasting inserted text in Visual mode does not work properly
8.0.0141 (after 8.0.0137) nested function test fails on AppVeyor
8.0.0142 normal colors are wrong with 'termguicolors'
8.0.0143 line number of current buffer in getbufinfo() is wrong
8.0.0144 when using MSVC the GvimExt directory is cleaned twice
8.0.0145 running tests on MS-Windows is noisy
8.0.0146 termguicolors uses wrong colors on MS-Windows with ConEmu
8.0.0147 searchpair() fails when 'magic' is off
8.0.0148 wrong indent in C preprocessor with line continuation
8.0.0149 :earlier does not work after reading the undo file
8.0.0150 completion for :filter does not skip the pattern
8.0.0151 passing buffer content to system() is clumsy
8.0.0152 stray channellog created when running tests
8.0.0153 (after 8.0.0151) system() test fails on MS-Windows
8.0.0154 (after 8.0.0151) system() test fails on OS/X
8.0.0155 ubsan complains about NULL pointer
8.0.0156 not enough test coverage for float functions
8.0.0157 no completion for :syntax spell and :syntax sync
8.0.0158 (after 8.0.0156) float function test fails on MS-Windows
8.0.0159 crash on startup when updating tabline
8.0.0160 EMSG() is sometimes used where it should be IEMSG()
8.0.0161 (after 8.0.0159) can't build with small features
than one encoding matches, by comparing with 'and bitmask' rather than
equality to bitmask.
Bump PKGREVISION for functional change.
Patch from Yasuhito FUTATSUKI in PR pkg/51792
Not needed any longer, tested with emacs25-nox11 on monster; others
copy look quite the same.
I suspect this is not needed on FreeBSD, OpenBSD, or Linux either, but
don't have test hosts.
- set ELFSIZE correctly (I still have no idea how this ever worked)
- insert workaround for gcc5's calloc bug
For some reason I thought emacs22 wasn't affected by these issues, or
I would have done this earlier.
This does not make the package build on -current, but it gets a lot
farther now - it segvs building the quail elisp. I have a feeling
we've seen this problem before also, but if so it was some years back
and I can't remember what it was about.
These were release candidates according to upstream (Stephen
Turnbull):
"Right now, everything has gotten all the beta testing it's going to
get, so you can just package it from experimental."
<http://xemacs-beta.xemacs.narkive.com/TgII9kK7/when-do-experimental-packages-turn-stable>
but did not make to final before xemacs.org went south.
In addition, teach the update mechanism to use a configurable ftp
server.
2016.12.12 - GNU nano 2.7.2 "Shemesh! Shemesh!" brings another feature:
the ability to complete with one keystroke (^] by default)
a fragment of a word to a full word existing elsewhere in
the current buffer. Besides, this release fixes two bugs
related to using line numbers in softwrap mode, allows to
use the PageUp and PageDown keys together with Shift on
VTE-based terminals, stops the help lines from flickering
during interactive replacing, makes a "set fill" override
an earlier "set nowrap", properly restores the selected
region after an external spell check, and improves a few
other tidbits. If you should find any more bugs, please
run 'man nano | grep bugs' and report them there.
I believe the segfault seen during building on DragonFly is a common one
seen by many platforms using GCC 5.x and later. It's been fixed on Emacs
25 though. Even though it's a generic GCC 5+ fix, limit it to DragonFly
only for now. Taken from DPorts (from archives)
The changes are very numerous. for the full list of changes in LyX 2.2.x,
refer to http://wiki.lyx.org/LyX/NewInLyX22
Short version of LyX 2.2 user-visible changes:
HiDPI display support
Qt5 support
New text display algorithm
Horizontal scrolling for large insets
New separators and improved control of paragraph breaks
Improved text color support
Improved box dialog
Improvements to tables
Reworked instant preview
Enhanced source panel
Improved outliner and navigation menu
Enhanced language support
New supported languages
New supported LaTeX commands
Layout enhancements
Module enhancements
New preferences settings
New converters
that if you dump with a non-ASLR temacs you get a working emacs binary, and
if you don't you don't, although I don't really see why -- perhaps it's
something broken in crtstuff. Closes PR 51654.
Note that pre-ASLR emacs20 binaries not dumped by an ASLR temacs also
blow up in the same way, which doesn't make much sense either, but
undoubtedly it's all connected.
It's not particularly good that we apparently don't have backwards
compatibility for old Emacs binaries because of this, but for the time
being I'm more worried about it working at all.
PKGREVISION++ again, to 22.
