Re: Input line editing

[christos%astron.com@localhost (Christos Zoulas)]

In article <20090213142812.3d004b14%hal.xisop@localhost>,
Matthew Mondor  <mm_lists%pulsar-zone.net@localhost> wrote:
>On Fri, 13 Feb 2009 11:36:06 -0500 (EST)
>der Mouse <mouse%Rodents-Montreal.ORG@localhost> wrote:
>
>> > Add paging (optional) to the processing.
>> 
>> Hmm, worth thinking about.  It would be relatively intrusive, as input
>> line editing currently has nothing to do with output processing except
>> for noticing when output happens (to drive auto-retype).
>> 
>> One thing I do not want is for output to have to bounce through the
>> line editor process on the way out.  While ttys with ICANON set are not
>> used for bulk data input (see upthread), they _are_ used for bulk data
>> output, so the performance issue is no longer an automatic non-issue.

I thought about doing command line editing in the kernel as a line discipline
in the late 80's. There are many problems one needs to solve:
        - key binding
        - passing termcap info to the kernel for multi-line stuff
        - dealing with history for separate programs
        - complex code handling incremental changes to avoid full redraws
        - providing completion
        - dealing with automatic and magic margins
        - handing internationalization and double width characters

It is really not worth the trouble

christos