Browse Source

Redirect readers of CHANGES and ISSUES to the actual web pages

git-svn-id: https://klish.googlecode.com/svn/trunk@597 0eaa4687-2ee9-07dd-09d9-bcdd2d2dd5fb
Serj Kalichev 12 years ago
parent
commit
cc624ef42c
2 changed files with 3 additions and 672 deletions
  1. 2 581
      CHANGES
  2. 1 91
      ISSUES

+ 2 - 581
CHANGES

@@ -1,581 +1,2 @@
-/**
-\page _changes_page Changes
-
-<HR>
-
-    Serj Kalichev
-
-\section _1_5_0 1.5.0
-Thanks to the contributors:
-- Michal Turek
-- Camilo A. Arboleda, Cybercom
-
-\section _1_2_1 1.2.1
-
-\section _1_2_0 1.2.0
-
-\section _1_1_0 1.1.0
-
-\section _1_0_0 1.0.0
-- The project was forked. The new name is klish.
-
-The following updates were introduced as part of work undertaken as an employee
-of Newport Networks. (http://www.newport-networks.com)
-
-    Graeme McKerrell
-
-\section _0_7_3 0.7.3
-- minor bug fix from Cliff Martin which avoids a SEGV when CTRL-C is invoked
-  on an empty line.
-
-\section _0_7_2 0.7.2
-- updated the default value fix slightly to ensure that if the default is set
-  to the empty string then the parameter will not be translated. This means
-  that legacy code which uses an empty default as an indicator of a parameter
-  not being provided will continue to work.
-  
-- incorporated bug from Charlie (nmsguru at sourceforge dot net) to make
-  sure that default values are correctly translated. e.g. a SELECT parameter
-  can be given the textual name in the XML and will be translated into the
-  appropriate interpretation when used.
-
-- updated the configuration so that the --enable-lubheap option is disabled by default. There
-  were a number of people which were having problems compiling the memory management code and tests.
-
-- added in lub_partition component which provides a fast local memory heap for the current thread
-  as well as a global mutex controlled heap which is used by multiple threads.
-  
-- added statistics to lub_blockpool component
-
-- performance overhaul of lub_heap component. Added a cache for smaller sized blocks which provides
-  a boost to the alocation/deallocation speed.
-  
-- updated escape_decode() function in vt100.c to become non-blocking for
-  the scope of decoding an escape sequence. This prevents hanging the
-  input if a user presses the ESC key.
-  
-- added a lub_heap_show_allocs() function which enables any leak detection
-  tracked allocations to be output. This is useful in the case where memory
-  corruption is preventing a scan from taking place.
-  
-- updated lub_heap__set_framecount() so that when you disable leak detection
-  the tainting state is returned to that before leak detection was enabled.
-
-- implemented correct alignement allocations of sizes other than 
-  native alignment. Changed the naming convention to 
-  LUB_HEAP_ALIGN_2_POWER_X and provided support upto 128MB alignment.
-  
-- updated the memInit() function in lubheap/vxwork/memLib.c to call
-  memLibInit() and memPartLibInit(). This is necessary for systems which
-  only call memInit() in usrRoot() routine.
-
-- updated the root-view.xml example file to demonstrate the use of different
-  parameter types.
-
-- updated the named-view.xml example file to demonstrate the use of viewid.
-
-- added patch from Anselmo Lacerda Silveira de Melo to add CTRL-U key
-  combination to tinyrl which deletes the current line contents.
-
-- fixed "const-ness" issue in lub/heap/context.c
-
-- added error checking to posix/heap_leak_mutex.c in case the pthread
-  calls fail.
-  
-- added heap address to the details output by lub_heap_show()
-
-- fixed possible uninitialised variable in heap_realloc.c
-
-- updated clish.xsd to make sure the "PARAM" tag is named.
-
-\section _0_7_1 0.7.1
--  Cleaned up configure.ac, added --enable-lubheap experimental option which can
-   be used to replace the normal system memory libraries with a lub/heap implementation.
-   Added librt and libintl support as well as POSIXifiying the build for clean
-   building under LINUX.
-
--  Updated the shell_spawn.c file to enable tilde expansion in CLISH_PATH.
-
--  split lub_heap platform specifics into two parts 
-   1) the platform specific support code required by the core 
-      lub_heap implementation. 
-   2) a lubheap component which can be linked in to replace the normal
-      platform memory libraries.
-
--  updated lub_heap component so that memory scanning is handled by specific functions
-   rather than assume the naming conventions of particular sections during the link.
-   The platform specific code now has to perform the scanning.
-
--  refactored lub/heap/vxworks to better override vxworks libraries when linked in
-   as a library.
-
--  fixed lub_heap component so that it compiles on 64 bit architecture, and with GCC-4.0
-
--  updated tinyxml to version 2.5.1
-
--  updated lub/heap/vxworks/malloc.c/free.c/realloc.c to undefine any
-   MACRO set with their function name. This ensures that they always get compiled 
-   with the correct name, even if the rest of the CLISH framework is being 
-   compiled to use an alternative memory heap, via -D options
-   
--  Fixed issue in tinyrl_readline where raw mode was enabled (and input buffer
-   flushed) between printing prompt and accepting input. This was causing
-   things like expect scripts to fail. Fix was to change to raw mode prior to
-   printing the prompt.
-
--  Fixed lub_heap component to work with latest version of Cygwin.
-   The areas which use the __builtin_return_address() function,
-   provided by GCC, now uses the lub_heap_data_start value to
-   check whether a returned address is beyond the "text" section.
-   Hence a stack trace is deemed to come to an end if either the return
-   address is 0 or greater than lub_heap_data_start.
-   
-- Changed next pointer in lub_heap_s to be "lub_heap_s *"as opposed 
-  to "lub_heap_t *" so that GDB does not give incomplete type.
-
-  Fixed memory stomp in tinyrl_completion.
-
-- Updated to account for obscure problem related to C-strings
-  being signed and the ctype.h API taking a signed integer.
-
-  If a string contains characters greater than 127 then the signed bit
-  will be set and when calling a ctype.h utility this can get sign
-  extended.
-  e.g. 
-  char *ptr = "£10"; isspace(*ptr);  will have the value -100
-  passed to the function (or MACRO) as the '£' character (decimal 156)
-  is negative when held as a signed char.
-
-  The solution is to ensure that you always used unsigned chars to
-  pass to the utility functions. This was achieved by adding a lub_ctype 
-  component which provides 8-bit safe versions of these functions.
-
-- added ability to have scripts stop on error or not by introducing a new
-  clish_source_no_stop entry point,
-
-- added ability to limit the maximum line length in tinyrl.
-
-- fixed bug in tinyrl_completion which was stomping on memory when there
-  were an exact multiple of 10 completions. Fix is to allocate one extra
-  entry for the NULL terminator.
-
-- added lub_heap_destroy() operation which frees up a heap object so
-  that the memory segments used for it can be reused for something else.
-  
-- Updated lub_heap so that leak scanning will account for
-  any non-monitored allocated blocks in the system. This 
-  prevents false leaks from being reported. However it
-  cannot account for any of the non-monitored blocks being
-  leaked themselves.
-
-- stripped out some spurious code in vxworks/heap_symShow.c
-
-- lubMallocTest.exe now exercises stress test on lub_heap component.
-  mallocTest.exe runs the same test on the native heap manager for the
-  host platform.
-
-- fixed lub_heap__get_max_free() to work when no free blocks are present.
-
-- added a generic lub_heap_scan_stack() function which can be used
-  to scan any variables held in the current stack up to 32 levels.
-  (needed for non VxWorks systems)
-
-- made lub_heap_scan_memory() take a const void* argument.
-
-- updated lub/heap.h to work when included from C++ file.
-
-- updated the posix/heap_symShow.c to use libbfd for symbol decoding.
-  A lub_heap_init() operation now exists which is called to
-  provide the filename of the current program so that symbols can be
-  loaded.
-
-- disabled memory leak detection and memory taiting by default for
-  the POSIX platform.
-
-- changed default behaviour to tainting and leak detection off.
-
-- when leak detection is enabled for the first time an exit handler is
-  registered to dump memory when a program completes.
-
-- updated configure.ac so that -disable-gpl may be used to prevent
-  linking with GPL licenced libraries.
-
-- fixed bug in tinyrl_completion which was stomping on memory when there
-  were an exact multiple of 10 completions. Fix is to allocate one extra
-  entry for the NULL terminator.
-
-- fixed bug in clish_pargv_new() which was only allocating enough memory 
-  for the number of arguments entered on the command-line. In the case of
-  optional trailing parameters, which were not entered, this caused a memory
-  stomp! Fixed by allocating enough memory for the maximum of the entered 
-  parameters or the command parameters.
-
-- added clish_command__get_param_count() method to obtain the number of 
-  parameters associated with a command.
-
-- fixed bug in lub_heap_graft_to_bottom; the tail was not getting it's 
-  free and segment fields set up.
-
-\section _0_7_0 0.7.0
-- Updated shell_execute.c to ensure that directories are not executed
-
-- fixed minor bug in autocompletion of parameters. 
-  clish_shell_parameter_generator() needed to account for trailing space 
-  when giving context sensitive help for parameters.
-  
-- updated lub_heap to maintain the length of blocks inclusive of the block
-  headers and tail. This reduces the overhead of manipulating sizes during
-  operations and also addresses an issue with the re-use of previously freed blocks.
-
-- added dblockpool component for creating dynamic chunks to be used as 
-  blockpools. The lub_heap implementation now uses this for holding
-  context instances, which both reduces the overhead memory used and
-  greatly reduces fragmentation when detecting leaks.
-  
-- fixed double insert bug in the lub_bintree_insert.c file.
-  Although the function was correctly returning -1 it was also 
-  replacing the root node with the provided client node!
-
-- updated shell_tinyrl.c so that typing "?" whilst in the middle of a 
-  quoted string inserts the '?' character rather than giving context
-  sensitive help.
-
-- fixed memory leaks which occurred when clish thread is cancelled during 
-  execution of a script.
-
-- updated to make the execution of a script asynchronously cancellable.
-
-- modified clish_shell's history command to use new methods 
-
-- added tinyrl_history_entry__get_index() method and removed 
-  tinyrl_history__get_base()
-
-- modified tinyrl_history so that index is held in the history_entry 
-  instances Then updated the history insertion code so that duplicate
-  entries are only held once in the history. This means that numeric
-  gaps can now occur in the history list, but a stifled list will no
-  longer force recent commands off the list through the repetition of 
-  a single command.
-  
-- updated tinyrl_completion() so that if the completion list contains
-  words which have a common (case insensitive) prefix are correctly
-  handled. Before this fix if commands "one" and "OneTwoThree" both
-  existed in the same view then the user would be unable to type
-  anything beyond "one".
-
-- fixed bug in clish_shell_tinyrl_key_enter() where a pointer to the line
-  for the current tinyrl instance was being held onto after it had been 
-  changed by an auto-completion, hence was now referencing freed memory.
-
-- ensure that when reading a script and an error occurs we don't bother
-  to show the errored command line after the "DING"
-
-- ensure that after reading a script the prompt isn't duplicated.
-
-- fixed memory leak in clish_pargv_init() where the 'args' parameter wasn't
-  releasing its working string.
-
-- fixed tinyrl_redisplay so that CTRL-K works; wasn't accounting for 
-  insertion point movements before erasing characters from end of line.
-
-- fixed history.c so that stifled history lists add_n_replace() properly.
-  This also addresses an assertion which was occuring with "history 1"
-
-- updated clish_shell class so that when an error occurs in a script, 
-  execution is terminated in any parent scripts. 
-  This doesn't terminate an interactive session.
-  This means that a script will now only run until it's first error 
-  (or the first error of a child script etc...)
-
-- plugged a poential memory leak in lub_string_catn() function.
-
-- fixed clish_shell_delete() so that the file stack is cleared out.
-
-- updated lub/argv class to be more resilient when memory
-  allocation failures occur.
-
-- reduced the terminal noise generated by tinyrl when 
-  auto-completion occurs.
-
-- reduced the termnal noise generated by tinyrl when user deletes
-  characters from the end of a line.
-   
-- VxWorks has an issue with its pthreads library which means that a
-  cleanup hook can be called multiple times (recursively)!!!. Updated
-  the cleanup function to set the pthread cancel state to 
-  PTHREAD_CANCEL_DISABLE to prevent this.
-  
-- found and fixed a memory leak in the stack of files being used for input.
-
-- updated auto-completion code to handle quoted arguments.
-
-- fixed echoing issue when echoing should be hidden.
-
-- overhauled the auto-completion code. Now capable of autocompleting 
-  parameters as well as commands. Parameters are now validated as they
-  are typed. This has involved some restructuring of the
-  tinyrl_completion interfaces.
-
-- fixed so that validation of integers works correctly.
-
-- removing the spurious auto-completions has had the side effect of not 
-  echoing commands executed from scripts. fixed this issue.
-  
-- modified the help system so that if a command has possible other
-  commands which are an extension of the current command name 
-  (e.g. slot, slotOne, slotTwo) then the command help will be augmented
-  the summary help of the possile completion commands.
-  
-- fixed bug where the SPACE key could cause spurious completion 
-  information to be displayed before entering a space character.
-
-- fixed clish_shell_help.c so that any detailed help held for
-  "place-holder" commands (e.g. "show slot" has a place holder command 
-  called "show") will be displayed when ?? is entered.
-
-- fixed bug where entering a command which happened to be a prefix
-  of other commands, caused spurious completion information to be
-  displayed before executing the command.
-
-- cleaned up the tinyrl.c to remove a spurious tinyrl_crlf() call
-  when reading a line. If a command outputs no text to stdout/stderr
-  then the prompt will now reappear on the following
-  line rather than leave a blank.
-
-- Updated the context sensitive help so that any parmeters which
-  have specific ranges e.g. "integer" or "selection" indicate
-  these possible values in the help text.
-  
-- Updated the ptype class to allow for pre-processing of parameter
-  before validation. See the XML schema for details.
-
-- Updated the ptype class to allow three methods for parameter validation.
- "regexp"[default], "integer" and "select" See the XML schema for details.
-
-- Went through the code cleaning up PC Lint issues.
-
-- Added a last_buffer field to the tinyrl class to cache the last
- displayed line. This can then be used to reduce the number of output
- characters when only a single char has been added to the end. 
- This simplifies the automated testing/interaction with the CLI.
- e.g. expect scripts.
- 
-- updated tinyrl.c to cache the isatty settings. This provides a
-  new operation called tinyrl__get_isatty()
-
-- removed nested shell support for a single client. It is no longer
-  needed as the "clish_source" command is used instead of spawning
-  another shell.
-
-- changed so that history entries are only added for interactive lines.
-  So if for example a user were to "clish_source" another file, 
-  the commands so issued from that file will not be added to the history.
-  
-- changed "clish_spawn" back to "clish_source" and implemented a stack of 
-  file handles per shell instance. The spawning of a sub-script was 
-  resource intensive, and had issues with spawned pthreads not having 
-  the attributes (e.g. stack size) as the parent thread. It also felt
-  too clunky to have multiple shell associated with a single client.
-
-- fixed tinyrl_readline() to make sure that lines longer than 80 chars are
-  handled correctly when read from a script.
-  
-- modified tinyrl_replace_line() to only modify the insertion point if
-  the buffer shrinks beyond the current insertion point. This
-  addresses the inserting spaces in the middle of a line issue.
-  
-- fixed clish_shell_spawn_from_file() so that it returns 
-  BOOL_TRUE upon success.
-
-- fixed tclish executable to read input from files correctly.
-
-- updated tinyrl.c to simply read non-tty input streams rather than
-  try and treat them like an interactive session.
-  
-- fixed tclish executable to account for nested shells being attached
-  to a single client. (using the "clish_spawn" builtin command.) 
-  Only when the root shell is initialised or finalised will
-  the resources be allocated and freed.
-
-- renamed the "clish_source" builtin command to "clish_spawn"; it takes
-  a filename and spawns a new shell to interpret the contents of the file.
-
-- added tinyrl__get_istream() tinyrl__get_ostream(), 
-  tinyrl_vt100__get_istream() and tinyrl_vt100__get_ostream() operations. 
-  This means that a client of CLISH can identify which file handles are 
-  being used. This is useful in the case of a client which is sourcing 
-  files using the "clish_source" builtin command.
-
-- Fixed to ensure that SPACE auto-completion works with commands which
-  take "args" type parameters.
-  
-\section _0_6_3 0.6.3
-- Updated tinyrl.c to ensure that SPACE auto-completes multi-word commands.
-  
-- Updated tinyrl.c to ensure that the line is redisplayed after a
-  completion has been performed. This is needed to ensure that when
-  a line is complete e.g. when pressing ENTER, the completion is 
-  shown before moving on.
-
-- fixed a freed memory usage in clish/shell/shell_tinyrl.c where a 
-  line pointer was obtained before completion was performed, then used. 
-  Needed to re-aquire the line pointer after completion, because it
-  could have changed.
-
-- fixed a memory corruption in tinyrl/history.c where a spurious NULL
-  terminator was being added to the vector of history entries
-  and corrupting the heap.
-
-- Updated command help to show the full multi-word command in the
-  detailed help. Also shows required parameters &lt;like_this&gt;
-
-\section _0_6_2 0.6.2
-- removed any non-interpreter specific setup/restore details from
-  tclish_init_callback and tclishell_fini_callback and put them into 
-  tclish.cpp
-  
-- updated tinyrl_readline() to change the terminal input mode on the fly 
-  to raw mode and then restore it when the line has been entered. This
-  is needed to ensure that the terminal is in the correct mode for any 
-  sub-shells which are launched. This means that clish_init_callback.c
-  and clish_fini_callback.c are redundant so have been removed.
-
-- updated clish_shell_spawn interface to take a pthread attibute as
-  a parameter this makes it possible for a client to control the
-  thread details.  
-
-- Documentation cleanups; added link to sourceforge pages, removed
-  redundant issues, started a doxygen todo list.
-
-\section _0_6_1 0.6.1
-- Fixed a bug where trying to insert text in the middle of a line
-  caused the remains of the line to be deleted.
-
-- updated the tdemo script to use the normal installation location for TCL
-  /usr/share/tcl8.4 rather than /usr/local/share/tcl8.4 which is used when 
-  you build it from scratch.
-
-\section _0_6_0 0.6.0
-
-    
-\subsection _0_6_0e 0.6.0e
-- Fixed shell_spawn.c so that client initialisation can fail hence cause the 
-  creation of the shell instance to be terminated. This means that client's 
-  may choose to perform user authentication at this point if they so wish.
-
-- Added a cmd_line hook which is called to provide the client details of 
-  each command line accepted by the shell. This enables logging of this
-  detail if so required. 
-
-\subsection _0_6_0d 0.6.0d
-- updated the autoconf system to use the TCL version for the tclishX.X name.
-
-- fixed clish/tclish_init_callback.c so that Tcl_FindExecutable() is called.
-  This is required to setup the TCL environment correctly. 0.6.0c release 
-  didn't load init.tcl correctly and was also crashing when file operations
-  were performed (e.g. glob, pwd)
-
-- finalized clish.xsd with proper namespace details
-
-
-\subsection _0_6_0c 0.6.0c
-- Fixed autoconf scripts to check for pthreads and TCL library.
-
-- Updated lub_argv class to be able to tell you when a parameter was quoted.
-  The paramter value itself will not contain the " marks.
-  
-- Introduced "args" and "args_help" attributes to the COMMAND element. 
-  This enables the remenant of a command line, after the parameters defined in
-  the PARAMS elements, to be assigned to a psuedo parameter variable.
-
-- Introduced a "escape_chars" attribute to the COMMAND element. This allows 
-  commands to override the default selection of characters which are escaped 
-  (e.g. \$) when expanded from a variable. Escaping of special characters is a
-  security feature to prevent the user from executing an underlying shell
-  command through the use of "clever" parameters. e.g. "fred`rm -rf \*`bloggs"
-  would be a rather unfortunate parameter on a UNIX system!!!
-
-- Introduced a "builtin" attribute to the ACTION element. Also enabled clients
-  to register their own built commands when spawning a shell. The XML MUST now
-  specify a builtin attribute to invoke a builtin command callback.
-  
-- entering an empty line simply redisplays the prompt. This enables a user to 
-  create some clear space on the screen.
-
-- Added tinyxml into the distibution; currently the tarball of this 
-  component (available from http://www.grinninglizard.com/tinyxml/index.html)
-  doesn't support autoconf, so by adding this in (plus a module.am) you get 
-  autoconf support. This library is 'much' smaller than libxml2 and hence is
-  more appropriate for an embedded system. libxml2 is no longer needed by 
-  clish.
-  
-- Added spawn interface to the shell class. This causes a separate thread to 
-  be lauched to handle the shell session. This is required for flat namespace 
-  embedded envrioments where you may need to sustain multiple concurrent
-  sessions.
-  
-- Introduced tinyrl component. This is a simplistic replacement for the 
-  readline library, which cannot be used in embedded-single memory space 
-  systems due to the use of global variables. Tinyrl implements sufficient 
-  functionality for the clish component to function, and no-more although more 
-  may be added in the future. readline is no longer needed to build clish.
-
-- Updated clish component so that "viewid" tag can be used to define view
-  specific variables. The syntax is of the form "fred=hello world;bert=goodbye"
-  and causes ${fred} and ${bert} to expand to the assigned values whilst in the
-  new view.
-  NB. the values themselves may be based on variables available in the current
-  context (e.g. environment, parameters or current viewid variables.)
-  
-- Updated clish component to use history escape sequences via an external
-  history_expand function. 
-  tinyrl_history currently supports "!!" "!N" and "!-N" expansion.
-  
-- Updated clish component to not use the exit(1) command to terminate the shell
-  instead a "close" instruction can be given to the shell and it will exit it's
-  external loop and terminate normally. This is for ease of use in 
-  single-memory-space embedded systems.
-  
-- Added a partial string test, which confirmed a bug in the VxWorks standard
-  headers which was causing lub_string_nocase_cmp() to work incorrectly.
-  
-- Updated clish component to support STARTUP XML element. This is used to 
-  display a banner and perform a startup action when a shell is first started.
-
-- Updated clish component to support OVERVIEW XML element. This is used to 
-  display some textual help when the clish_overview internal command is 
-  specified in an ACTION element.
-  
-- Updated clish component to support the DETAILS XML element.
-  These enhance the context sensitive help with the ability to provide more
-  details for a particular command. (a repeated press of the '?' key displays
-  this extra detail)
-  
-- Introduces more built in commands (can be specified in the ACTION element)
-  There are now four in total.
-  "clish_close"        - terminates the session.
-  "clish_overview"     - displays the text defined in the OVERVIEW element
-                         of the XML.
-  "clish_history [n]"  - displays the currently held history of command lines.
-                         Also enables the stifling of the history list to a
-                         value specified by the optional parameter 'n'. 
-                         A value of zero means unstifled.                   
-  "clish_source file"  - spawn a separate clish session reading the input from
-                         the specified file. Output still goes to the current
-                         output stream.
-
-- Mapped [space] and [enter] to auto-complete commands if appropriate.
-
-
-<HR>
-
-\section _0_5_4 0.5.4
-This work was done as an employee of 3Com (http://www.3com.com)
-
-    Graeme McKerrell
-
-- Initial public release
-
-
-*/
-*/
+See the http://code.google.com/p/klish/source/list for the changes of SVN repository.
+See the http://klish-cli.blogspot.com/ for the release changes and another klish related news.

+ 1 - 91
ISSUES

@@ -1,91 +1 @@
-/**
-\page _issues_page Issues
-
-\todo
-- placing a "?" character in a text file doesn't display help text when 
-  passed into clish.
-
-- ACTIONS which are meant to handle signals like CTRL-C 
-  don't seem to work properly, either the sub-script fails and the
-  shell breaks, or the sub-script never handles the signal (depending on 
-  platform).
-  
-- the heap unittest current has 10 failures
-
-- no unittests for the dblockpool component yet...
-
-- lub_heap_destroy() method should be provided which will de-register any allocated blocks from the
-  leak detection trees.
-
-- update leak detection to scan non-monitored allocated memory blocks, for references. This requires
-  the ability to iterate all the heaps in the system.
-
-- auto-completion currently doesn't handle optional parameters (defined using the "prefix" attribute).
-  
-- If a multi-word command is defined in global and local views then
-auto-completion currently works for both, but resolving doesn't seem to.
-\verbatim
-e.g. "mode"     - local view
-     "mode qos" - global view
-     Will auto-complete "mode qos" but won't run it...
-\endverbatim
-- Doesn't handle wrapping of long lines properly; get multiple lines scrolling
-  down the screen.
-- Not all files have been split out into one file per function...
-- need to complete the Doxygen documentation for clish, tinyrl and lub/argv
-
-\section _nice Nice to have
-- when importing from a namespace, allow the specification of the following:
-  no-help         - defaults to false but may be set true to prevent display in the help menu.
-  no-completion   - defaults to false but may be set true to prevent auto-completion. 
-  no-context-help - defaults to false but may be set true to prevent context sensitive help for 
-                    commands.
-
-- place an access level on namespaces; saves worrying about defining access for
-  every single command. e.g. device_show namespace could have different access
-  to device_config namespace. device view simply imports both namespaces.
-
-- When defining a default attribute if you could include a variable in the 
-  definition, default could be given meaningful values.
-- Augment the XML schema for support of dynamic variables.
-  e.g. 
-  \verbatim
-  <VARIABLE name="fred"><ACTION>puts "hello world"</ACTION></VARIABLE>
-  \endverbatim
-  will get the value of "hello world" when expanded as ${fred}... 
-  This is simple for TCL as the result is returned... what about for clish ,  
-  how do we get a return value into the execitable?
-
-- Currently handles commands in one view or all views. Better model is to be able
-  to specify a namespace for commands and then assign a view to a namespace.
-  e.g.
-  \verbatim
- <NAMESPACE name="fred">
-    <COMMAND>
- </NAMESPACE>
- <VIEW name=" kdjkdj"...>
-    <IMPORT namespace="fred">
-    <COMMAND ...>
- </VIEW>
-  \endverbatim
-  by default a view imports commands from the global namespace.
-  NB. need to be able to resolve commands from all included namespaces.
-  
-- In addition a command could become an alias for an entire namespace's worth
-  of sub-commands.
-  e.g.
-  \verbatim
- <NAMESPACE name="config">
-   <COMMAND name="show"...>
-   <COMMAND name="show ip"...>
-   ...
- </NAMESPACE>
-  
- <VIEW name="foobar"...>
-    <COMMAND      name="do" 
-                  help="perform a configuration command" 
-             namespace="config"/>
- </VIEW>
-  \endverbatim
-
-*/
+See the http://code.google.com/p/klish/issues/list for the list of current issues.