Changelog for readline-devel-7.0-1.48.i586.rpm :

* Wed Jun 13 2018 wernerAATTsuse.de- Add patch readline-7.0-screen.patch to be able to parse settings in inputrc for all screen TERM variables starting with \"screen.\" to fix boo#1095661
* Sun Mar 18 2018 jengelhAATTinai.de- Update descriptions. Replace old $RPM_
* vars with macros.
* Fri Mar 16 2018 wernerAATTsuse.de- Apply the upstream patches as well!!!
* Fri Mar 16 2018 wernerAATTsuse.de- Avoid self Provides/Obsoletes
* Fri Mar 16 2018 wernerAATTsuse.de- Rename basic package to readline to have the latest readline version always within package readline again.
* Fri Mar 16 2018 wernerAATTsuse.de- Correct documentation path
* Wed Mar 14 2018 wernerAATTsuse.de- The obsoletes must be <=, as version stayed at 7.0
* Mon Mar 12 2018 schwabAATTsuse.de- Create separate readline7 package
* Tue Feb 27 2018 wernerAATTsuse.de- Create readline-devel-static package to re-enable static libraries again (boo#1082913)
* Thu Feb 22 2018 fvogtAATTsuse.com- Use %license (boo#1082318)
* Tue Feb 06 2018 avindraAATTopensuse.org- Add patch 19 to bash-4.4-patches.tar.bz2
* With certain values for PS1, especially those that wrap onto three or more lines, readline will miscalculate the number of invisible characters, leading to crashes and core dumps.
* Tue Jan 30 2018 avindraAATTopensuse.org- Add patches 13-18 to bash-4.4-patches.tar.bz2
* 13: If a here-document contains a command substitution, the command substitution can get access to the file descriptor used to write the here-document.
* 14: Under some circumstances, functions that return via the `return\' builtin do not clean up memory they allocated to keep track of FIFOs.
* 15: Process substitution can leak internal quoting to the parser in the invoked subshell.
* 16: Bash can perform trap processing while reading command substitution output instead of waiting until the command completes.
* 17: There is a memory leak when `read -e\' is used to read a line using readline.
* 18: Under certain circumstances (e.g., reading from /dev/zero), read(2) will not return -1 even when interrupted by a signal. The read builtin needs to check for signals in this case.- partial cleanup with spec-cleaner
* Wed Jan 24 2018 wernerAATTsuse.de- Modify patch bash-4.3-pathtemp.patch to avoid crash at full file system (boo#1076909)
* Fri Dec 08 2017 wernerAATTsuse.de- Enable multibyte characters by default
* Mon Sep 25 2017 wernerAATTsuse.de- Modify patch bash-4.4.dif to let bashline.h install as well as this header file is included by general.h due to the same patch (boo#1060069)
* Thu May 25 2017 bwiedemannAATTsuse.com- Make build reproducible in spite of profile based optimizations (boo#1040589)
* Wed May 24 2017 bwiedemannAATTsuse.com- Allow to disable do_profiling in builds (related to boo#1040589)
* Wed Apr 26 2017 wernerAATTsuse.de- Simplify patch readline-5.2-conf.patch
* Tue Apr 25 2017 wernerAATTsuse.de- Do not throw info and manual pages away
* Fri Feb 17 2017 wernerAATTsuse.de- Remove bash-4.0-async-bnc523667.dif as this one is fixed (and was disabled and nobody had reported trouble)
* Mon Jan 30 2017 wernerAATTsuse.de- Add upstream patch readline70-002 which replace old one There is a race condition in add_history() that can be triggered by a fatal signal arriving between the time the history length is updated and the time the history list update is completed. A later attempt to reference an invalid history entry can cause a crash.- Add upstream patch readline70-003 Readline-7.0 uses pselect(2) to allow readline to handle signals that do not interrupt read(2), such as SIGALRM, before reading another character. The signal mask used in the pselect call did not take into account signals the calling application blocked before calling readline().
* Fri Jan 27 2017 wernerAATTsuse.de- Add upstream patch bash44-006 Out-of-range negative offsets to popd can cause the shell to crash attempting to free an invalid memory block.- Remove patch popd-offset-overflow.patch to use bash44-006- Add upstream patch bash44-007 When performing filename completion, bash dequotes the directory name being completed, which can result in match failures and potential unwanted expansion.- Duplicate bash44-007 as readline70-002 as it seems to be missed- Add upstream patch bash44-008 Under certain circumstances, bash will evaluate arithmetic expressions as part of reading an expression token even when evaluation is suppressed. This happens while evaluating a conditional expression and skipping over the failed branch of the expression.- Add upstream patch bash44-009 There is a race condition in add_history() that can be triggered by a fatal signal arriving between the time the history length is updated and the time the history list update is completed. A later attempt to reference an invalid history entry can cause a crash.- Add upstream patch bash44-010 Depending on compiler optimizations and behavior, the `read\' builtin may not save partial input when a timeout occurs.- Add upstream patch bash44-011 Subshells begun to run command and process substitutions may attempt to set the terminal\'s process group to an incorrect value if they receive a fatal signal. This depends on the behavior of the process that starts the shell.- Add upstream patch bash44-012 When -N is used, the input is not supposed to be split using $IFS, but leading and trailing IFS whitespace was still removed.
* Thu Jan 19 2017 wernerAATTsuse.de- Remove -L option on screen call dues API change, now we depend on environment variables only.
* Fri Dec 09 2016 mliskaAATTsuse.cz- Enable -fprofile-correction to cover misleading profile created due to terminating_signal which does not return.
* Mon Nov 28 2016 wernerAATTsuse.de- Add upstream patch popd-offset-overflow.patch to fix boo#1010845 CVE-2016-9401: bash: popd controlled free (Segmentation fault) Remark: this is a simple Segmentation fault, no security risk
* Thu Nov 17 2016 wernerAATTsuse.de- Add upstream patch bash44-001 Bash-4.4 changed the way the history list is initially allocated to reduce the number of reallocations and copies. Users who set HISTSIZE to a very large number to essentially unlimit the size of the history list will get memory allocation errors- Add upstream patch bash44-002 Bash-4.4 warns when discarding NUL bytes in command substitution output instead of silently dropping them. This patch changes the warnings from one per NUL byte encountered to one warning per command substitution.- Drop no-null-warning.patch as bash44-002 is official replacement- Add upstream patch bash44-003 Specially-crafted input, in this case an incomplete pathname expansion bracket expression containing an invalid collating symbol, can cause the shell to crash.- Add upstream patch bash44-004 There is a race condition that can result in bash referencing freed memory when freeing data associated with the last process substitution.- Add upstream patch bash44-005 Under certain circumstances, a simple command is optimized to eliminate a fork, resulting in an EXIT trap not being executed. (boo#1008459)- Add upstream patch readline70-001 Readline-7.0 changed the way the history list is initially allocated to reduce the number of reallocations and copies. Users who set the readline history-size variable to a very large number to essentially unlimit the size of the history list will get memory allocation errors
* Mon Oct 24 2016 schwabAATTsuse.de- no-null-warning.patch: Don\'t warn about null bytes in command substitution
* Tue Oct 04 2016 wernerAATTsuse.de- Avoid confusing library path
* Fri Sep 16 2016 wernerAATTsuse.de- Update bash 4.4 final
* Latest bug fixes since 4.4 rc2- Update readline 7.0 final
* Latest bug fixes since 7.0 rc2
* New application-callable function: rl_pending_signal(): returns the signal number of any signal readline has caught but not yet handled.
* New application-settable variable: rl_persistent_signal_handlers: if set to a non-zero value, readline will enable the readline-6.2 signal handler behavior in callback mode: handlers are installed when rl_callback_handler_install is called and removed removed when a complete line has been read.- Drop patch bash-4.3-async-bnc971410.dif as this one is part of 4.4- Drop patch bash-3.2-longjmp.dif as now long time be fixed- Drop patch bash-4.3-headers.dif as loadables now simply work- Drop readline-6.1-wrap.patch as this seems to be fixed- Disable patch bash-4.0-async-bnc523667.dif for now as it seems to be fixed in an other way
* Wed Sep 14 2016 wernerAATTsuse.de- Update bash 4.4 rc2 -- Bugfixes- Update readline 7.0 rc2 -- Bugfixes
* Mon Aug 01 2016 wernerAATTsuse.de- Make clear that the files /etc/profile as well as /etc/bash.bashrc may source other files as well even if the bash does not. Therefore modify patch bash-4.1-bash.bashrc.dif (bsc#959755)
* Thu Jul 14 2016 wernerAATTsuse.de- Update bash 4.4 beta 2
* Value conversions (arithmetic expansions, case modification, etc.) now happen when assigning elements of an array using compound assignment.
* There is a new option settable in config-top.h that makes multiple directory arguments to `cd\' a fatal error.
* Bash now uses mktemp() when creating internal temporary files; it produces a warning at build time on many Linux systems.- Update to readline library 7.0 beta 2 (not enabled as not standalone)
* The default binding for ^W in vi mode now uses word boundaries specified by Posix (vi-unix-word-rubout is bindable command name).
* rl_clear_visible_line: new application-callable function; clears all screen lines occupied by the current visible readline line.
* rl_tty_set_echoing: application-callable function that controls whether or not readline thinks it is echoing terminal output.
* Handle >| and strings of digits preceding and following redirection specifications as single tokens when tokenizing the line for history expansion.
* Fixed a bug with displaying completions when the prefix display length is greater than the length of the completions to be displayed.
* The :p history modifier now applies to the entire line, so any expansion specifying :p causes the line to be printed instead of expanded.
* Tue Mar 08 2016 wernerAATTsuse.de- Update bash 4.4 release candidate 1
* There is now a settable configuration #define that will cause the shell to exit if the shell is running setuid without the -p option and setuid to the real uid fails.
* Command and process substitutions now turn off the `-v\' option when executing, as other shells seem to do.
* The default value for the `checkhash\' shell option may now be set at compile time with a #define.
* The `mapfile\' builtin now has a -d option to use an arbitrary character as the record delimiter, and a -t option to strip the delimiter as supplied with -d.
* The maximum number of nested recursive calls to `eval\' is now settable in config-top.h; the default is no limit.
* The `-p\' option to declare and similar builtins will display attributes for named variables even when those variables have not been assigned values (which are technically unset).
* The maximum number of nested recursive calls to `source\' is now settable in config-top.h; the default is no limit.
* All builtin commands recognize the `--help\' option and print a usage summary.
* Bash does not allow function names containing `/\' and `=\' to be exported.
* The `ulimit\' builtin has new -k (kqueues) and -P (pseudoterminals) options.
* The shell now allows `time ; othercommand\' to time null commands.
* There is a new `--enable-function-import\' configuration option to allow importing shell functions from the environment; import is enabled by default.
* `printf -v var \"\"\' will now set `var\' to the empty string, as if `var=\"\"\' had been executed.
* GLOBIGNORE, the pattern substitution word expansion, and programmable completion match filtering now honor the value of the `nocasematch\' option.
* There is a new ${parameterAATTspec} family of operators to transform the value of `parameter\'.
* Bash no longer attempts to perform compound assignment if a variable on the rhs of an assignment statement argument to `declare\' has the form of a compound assignment (e.g., w=\'(word)\' ; declare foo=$w); compound assignments are accepted if the variable was already declared as an array, but with a warning.
* The declare builtin no longer displays array variables using the compound assignment syntax with quotes; that will generate warnings when re-used as input, and isn\'t necessary.
* Executing the rhs of && and || will no longer cause the shell to fork if it\'s not necessary.
* The `local\' builtin takes a new argument: `-\', which will cause it to save and the single-letter shell options and restore their previous values at function return.
* `complete\' and `compgen\' have a new `-o nosort\' option, which forces readline to not sort the completion matches.
* Bash now allows waiting for the most recent process substitution, since it appears as $!.
* The `unset\' builtin now unsets a scalar variable if it is subscripted with a `0\', analogous to the ${var[0]} expansion.
* `set -i\' is no longer valid, as in other shells.
* BASH_SUBSHELL is now updated for process substitution and group commands in pipelines, and is available with the same value when running any exit trap.
* Bash now checks $INSIDE_EMACS as well as $EMACS when deciding whether or not bash is being run in a GNU Emacs shell window.
* Bash now treats SIGINT received when running a non-builtin command in a loop the way it has traditionally treated running a builtin command: running any trap handler and breaking out of the loop.
* New variable: EXECIGNORE; a colon-separate list of patterns that will cause matching filenames to be ignored when searching for commands.
* Aliases whose value ends in a shell metacharacter now expand in a way to allow them to be `pasted\' to the next token, which can potentially change the meaning of a command (e.g., turning `&\' into `&&\').
* `make install\' now installs the example loadable builtins and a set of bash headers to use when developing new loadable builtins.
* `enable -f\' now attempts to call functions named BUILTIN_builtin_load when loading BUILTIN, and BUILTIN_builtin_unload when deleting it. This allows loadable builtins to run initialization and cleanup code.
* There is a new BASH_LOADABLES_PATH variable containing a list of directories where the `enable -f\' command looks for shared objects containing loadable builtins.
* The `complete_fullquote\' option to `shopt\' changes filename completion to quote all shell metacharacters in filenames and directory names.
* The `kill\' builtin now has a `-L\' option, equivalent to `-l\', for compatibility with Linux standalone versions of kill.
* BASH_COMPAT and FUNCNEST can be inherited and set from the shell\'s initial environment.
* inherit_errexit: a new `shopt\' option that, when set, causes command substitutions to inherit the -e option. By default, those subshells disable - e. It\'s enabled as part of turning on posix mode.
* New prompt string: PS0. Expanded and displayed by interactive shells after reading a complete command but before executing it.
* Interactive shells now behave as if SIGTSTP/SIGTTIN/SIGTTOU are set to SIG_DFL when the shell is started, so they are set to SIG_DFL in child processes.
* Posix-mode shells now allow double quotes to quote the history expansion character.
* OLDPWD can be inherited from the environment if it names a directory.
* Shells running as root no longer inherit PS4 from the environment, closing a security hole involving PS4 expansion performing command substitution.
* If executing an implicit `cd\' when the `autocd\' option is set, bash will now invoke a function named `cd\' if one exists before executing the `cd\' builtin.- Update to readline library 7.0 release candidate 1
* The history truncation code now uses the same error recovery mechansim as the history writing code, and restores the old version of the history file on error. The error recovery mechanism handles symlinked history files.
* There is a new bindable variable, `enable-bracketed-paste\', which enables support for a terminal\'s bracketed paste mode.
* The editing mode indicators can now be strings and are user-settable (new `emacs-mode-string\', `vi-cmd-mode-string\' and `vi-ins-mode-string\' variables). Mode strings can contain invisible character sequences. Setting mode strings to null strings restores the defaults.
* Prompt expansion adds the mode string to the last line of a multi-line prompt (one with embedded newlines).
* There is a new bindable variable, `colored-completion-prefix\', which, if set, causes the common prefix of a set of possible completions to be displayed in color.
* There is a new bindable command `vi-yank-pop\', a vi-mode version of emacs- mode yank-pop.
* The redisplay code underwent several efficiency improvements for multibyte locales.
* The insert-char function attempts to batch-insert all pending typeahead that maps to self-insert, as long as it is coming from the terminal.
* rl_callback_sigcleanup: a new application function that can clean up and unset any state set by readline\'s callback mode. Intended to be used after a signal.
* If an incremental search string has its last character removed with DEL, the resulting empty search string no longer matches the previous line.
* If readline reads a history file that begins with `#\' (or the value of the history comment character) and has enabled history timestamps, the history entries are assumed to be delimited by timestamps. This allows multi-line history entries.
* Readline now throws an error if it parses a key binding without a terminating `:\' or whitespace.- Remove patches which are upstream solved bash-3.2-longjmp.dif bash-4.3-headers.dif readline-6.1-wrap.patch- Rename patches bash-4.3.dif become bash-4.4.dif readline-6.3.dif become readline-7.0.dif- Refresh other patches as well
* Mon Oct 19 2015 wernerAATTsuse.de- Define the USE_MKTEMP and USE_MKSTEMP cpp macros as the implementation is already there.- Add patch bash-4.3-pathtemp.patch to allow root to clear the file systems. Otherwise the completion does not work if /tmp if full (ENOSPC for here documents)
* Fri Oct 16 2015 wernerAATTsuse.de- Remove --hash-size options as there is no any change in the final binary nor library anymore
* Mon Aug 31 2015 wernerAATTsuse.de- Add upstream patch bash43-039 Using the output of `declare -p\' when run in a function can result in variables that are invisible to `declare -p\'. This problem occurs when an assignment builtin such as `declare\' receives a quoted compound array assignment as one of its arguments.- Add upstream patch bash43-040 There is a memory leak that occurs when bash expands an array reference on the rhs of an assignment statement.- Add upstream patch bash43-041 There are several out-of-bounds read errors that occur when completing command lines where assignment statements appear before the command name. The first two appear only when programmable completion is enabled; the last one only happens when listing possible completions.- Add upstream patch bash43-042 There is a problem when parsing command substitutions containing `case\' commands within pipelines that causes the parser to not correctly identify the end of the command substitution.
* Wed Jul 01 2015 cooloAATTsuse.com- add bash-4.3-perl522.patch to fix texi2html for perl 5.22 (defined(AATTarray) has been deprecated since at least 2012)
* Thu May 28 2015 wernerAATTsuse.de- Add upstream patch bash43-034 If neither the -f nor -v options is supplied to unset, and a name argument is found to be a function and unset, subsequent name arguments are not treated as variables before attempting to unset a function by that name.- Add upstream patch bash43-035 A locale with a long name can trigger a buffer overflow and core dump. This applies on systems that do not have locale_charset in libc, are not using GNU libiconv, and are not using the libintl that ships with bash in lib/intl.- Add upstream patch bash43-036 When evaluating and setting integer variables, and the assignment fails to create a variable (for example, when performing an operation on an array variable with an invalid subscript), bash attempts to dereference a null pointer, causing a segmentation violation.- Add upstream patch bash43-037 If an associative array uses `AATT\' or `
*\' as a subscript, `declare -p\' produces output that cannot be reused as input.- Add upstream patch bash43-038 There are a number of instances where `time\' is not recognized as a reserved word when the shell grammar says it should be.
* Mon May 18 2015 meissnerAATTsuse.com- move info deletion to %preun sections
* Wed Mar 04 2015 jdelvareAATTsuse.de- bash-4.3-loadables.dif: One more warning fixed, in examples/loadables/logname.c.- bash-4.3-loadables.dif: Reverted one warning fix, which was introducing another warning and possibly a bug.
* Wed Mar 04 2015 jdelvareAATTsuse.de- bash-4.3-loadables.dif: Split changes to shell.h to a separate patch \"bash-4.3-include-unistd.dif\", as the loadables build just fine without these changes.- bash-4.3-loadables.dif: Drop all header file inclusion fixups, upstream fixed the problem differently 5 years ago.
* Wed Feb 18 2015 wernerAATTsuse.de- Do not restart all signal handlers for bash 4.3 as this breaks trap handler in subshells waotiug for a process
* Fri Jan 16 2015 wernerAATTsuse.de- Remove -DMUST_UNBLOCK_CHLD(=1) as this breaks waitchild(2) on linux
* Fri Jan 09 2015 wernerAATTsuse.de- Add upstream patch bash43-031 The new nameref assignment functionality introduced in bash-4.3 did not perform enough validation on the variable value and would create variables with invalid names.- Add upstream patch bash43-032 When bash is running in Posix mode, it allows signals -- including SIGCHLD -- to interrupt the `wait\' builtin, as Posix requires. However, the interrupt causes bash to not run a SIGCHLD trap for all exited children. This patch fixes the issue and restores the documented behavior in Posix mode.- Add upstream patch bash43-033 Bash does not clean up the terminal state in all cases where bash or readline modifies it and bash is subsequently terminated by a fatal signal. This happens when the `read\' builtin modifies the terminal settings, both when readline is active and when it is not. It occurs most often when a script installs a trap that exits on a signal without re-sending the signal to itself.
* Wed Dec 03 2014 jdelvareAATTsuse.de- Fix the sed command that fixes up the patch headers. It was printing a duplicate header line, which suprisingly did not confuse patch, but could in the future.- Fix all patches that had the duplicate header line issue.
* Tue Nov 04 2014 wernerAATTsuse.de- Use tail command to follow run-tests instead of a simpe cat command
* Fri Oct 24 2014 wernerAATTsuse.de- Really remove obsolete patches
* Fri Oct 24 2014 brianAATTaljex.com- Skip autoconf on OS 10.2 or older
* Wed Oct 22 2014 wernerAATTsuse.de- Avoid fdupes on SLES-10
* Wed Oct 22 2014 wernerAATTsuse.de- Bump bash version to 4.3
* Tue Oct 21 2014 brianAATTaljex.com- Allow building on targets from SL 10.1 to current since it\'s free