[Info-vax] VMS Features I Wish Linux Had

[Johnny Billquist]

On 2016-06-16 11:57, VAXman- at SendSpamHere.ORG wrote:
> In article <njtqt3$r4l$1 at Iltempo.Update.UU.SE>, Johnny Billquist <bqt at softjar.se> writes:
>> On 2016-06-15 20:26, Chris Scheers wrote:
>>> Stephen Hoffman wrote:
>>>> On 2016-06-15 11:57:06 +0000,   VAXman-  @SendSpamHere.ORG said:
>>>>
>>>>> THere IS line editing in the terminal driver but it's limited.
>>>>> Extending that to 'vi' and 'emacs' capabilities is NOT appropriate in
>>>>> the terminal driver.  Put it where it belongs!
>>>>
>>>> What most folks would want is some flexibility around editing command
>>>> input, allowing vim or emacs or (since this is OpenVMS) EDT-compatible
>>>> cursor control.
>>>>
>>>> http://www.catonmat.net/blog/bash-vi-editing-mode-cheat-sheet/
>>>>
>>>> The cursor handling is inherently going to be part of the driver (and
>>>> also sensitive to the character encoding), whether the command history
>>>> or recall buffer or the rest is part of the driver or is implemented
>>>> elsewhere is an implementation detail.  If OpenVMS ever went onto Mach
>>>> (past that prototype from many years ago), this handling could be in a
>>>> process somewhere, for instance, and for all anybody using it cared.
>>>>
>>>> Akin to the lack of FUSE support within the OpenVMS I/O and file
>>>> system implementation on OpenVMS, there's also no input character
>>>> handling layer on OpenVMS.
>>>
>>> It sounds like what is being asked for is something like an ACP (class
>>> driver?) for character I/O.
>>>
>>> Let the low level driver do driver stuff and move the bytes between the
>>> hardware and the machine.
>>>
>>> Let the "ACP" layer handle editing and formatting and provide
>>> consistency to all programs.
>>>
>>> Many "SET TERMINAL" settings would be handled in the "ACP".  You could
>>> even add settings to specify "vi", "emacs", "VMS", etc. line editing.
>>>
>>> I think this is what Steve means by "input character handling layer",
>>> but I would extend it beyond input to also include output formatting,
>>> command history, etc.
>>
>> Yes! That is a good concept. And in fact, this is what RSX has, and
>> there it is called an ACD - Ancilliary Control Driver. It hooks into the
>> terminal driver at certain points in the processing for both input and
>> output.
>>
>> The interface is not as clean as an ACP, as this is not something that
>> can be handled at the QIO system call level, but is processing that
>> needs to happen for individual characters as they are entered, or
>> strings when they are output, which can happen for other reasons than a
>> QIO as well.
>
> $ MCR TFU HELP ABOUT_TFF
>
>                    Terminal Fallback Facility (TFF)
>
>     The  VMS  Terminal  Fallback  Facility  (TFF)  provides table-driven
>     character   conversion   for   terminals.  The  conversion  is  done
>     transparent   to  the  application  (unnoticed  by  the  application
>     software  unless  explicit  inquiries  are  made).  You  can  choose
>     character conversion tables from a master character library, located
>     on the VMS system disk in:
>
>     SYS$SYSTEM:TFF$MASTER.DAT
>
>     This  library  contains  all  the character tables supported by TFF.
>     Conversions  can  be  one character to many on output to a terminal,
>     and one to one on input from a terminal.
>
>     TFF   can   also  perform  character  compose  emulation  on  input.
>     Furthermore, you can control which keys are auto-compose keys.
>
>     One  of  the  applications  of  TFF  is to allow users with National
>     Replacement  Character (NRC) set terminals to use software developed
>     with   the   Digital  Multinational  Character  Set  (MCS).  MCS  is
>     essentially  the  ASCII  character set plus 128 characters currently
>     used by owners of NRC terminals around the world.
>
>     In addition,  TFF  also  allows  users  to  use 16-bit character set
>     terminals,  such as  DOOSAN  Code  and  Telex  Code  Asian  language
>     terminals, for software developed  with  the Asian Language National
>     Standard Character Set.

That sounds really promising, except for one detail. The lack of one to 
many byte translations on input.

And I guess it's just a translation function, and cannot run any code at 
those points in time. Which probably would be needed if you wanted to 
implement something more complex, such as line editing.

	Johnny