On Tue, 6 Jun 2000, Jarom=EDr Dole=E8ek wrote:

>Tim Walls wrote:
>> > If their are no significant issues raised by Saturday night (11JUN2000
>> > 0000Z) -- that is, PRs giving detailed information on situations where
>> > pcvt works and wscons does not, hearsay and vague comments will not be
>> > accepted -- pcvt will be removed from the i386 port permanently.
>>=20
>> This is a conspiracy, isn't it?  First we implement the bastard son of
>> runlevels, the rape of rc.conf, and now they are nicking the console
>> driver I've used for years...
>>=20
>> It's part of the bigger plan - to remove practically everything I like
>> about NetBSD...  (paranoid?  moi?)
>
>Nono, it's not conspiracy :) It's just kind of christmas cleaning -
>throwing out old trash.
>
>pcvt was superceded by wscons, which is the mainline supported
>MI console driver. Supporting and maintaining two separate console
>drivers (furthermore if one of them is i386-specific) is just
>unnecessary and costs developer resources which can be more efficiently
>spent otherwise.
>
>Is there any feature pcvt has and wscons does not ?

Ability to map different escape sequences to different shifts of function
and cursor keys (as you can with xterm).  When I started using NetBSD a
couple of years ago it was quite easy (~1 evening) to modify pcvt to do
this by adding extra shift states to the key definitions table.  It lets
met get a set of key bindings (HP-PCterm) which I've used on every unix
system I've worked on over the past 10 years.

When wscons appeared I had a look at it and found that the code would need
major surgery to do the same thing because on those keys where some shifts
are recognised a different keysym is produced and the shift state is not
passed through to the function which defines the output strings.  The
fundamental problem is that the core wscons code makes assumptions about
which shift states will be recognised on which keys, rather than passing a
keysym and shift state to the terminal personality code so that it can do
what it wants.

Who "owns" the wscons code?  I'd like to discuss how I might make the
necessary changes in a way which could be accepted into the tree, and
doesn't conflict with anything anyone else may be doing.  For instance,
would it be better to do something which is compatible with the linux
console (where the string emitted by every shift combination on every
key can be changed dynamically)?


Roger

---------------------------------------------------------------------------=
---
Roger Brooks (Systems Programmer),          |  Email: R.S.Brooks@liv.ac.uk
Computing Services Dept,                    |  Tel:   +44 151 794 4441
The University of Liverpool,                |  Fax:   +44 151 794 4442
PO Box 147, Liverpool L69 3BX, UK           |=20
---------------------------------------------------------------------------=
---