Re: CONFIG_PNP: Please change the name

Andrew Mileski (dmtech@magi.com)
Sun, 17 Mar 1996 17:22:20 -0500 (EST)


>Not trying to discourage anybody doing development for Linux, this
>is one area where great care must be observed.
>The Notorious Plughg-and-Pray system. IMHO its useless in its best
>and a total disaster on its worst.
>If such a system comes to Linux, PLEASE make sure that in can be
>TOTALLY disabled.
>Cheers..
>--
>Lauri Tischler, Network Admin Tel: +358-0-47846331
>02211 Espoo FINLAND Internet: ltischler@fipower.fi

First, if you have a Pentium computer, you *ALREADY* have
one form of Plug-and-Play - it's called PCI!

*PLEASE* Don't compare the Windows95 or DOS lame-brained Plug-and-Play
configuration _ATTEMPT_ on this LINUX (a real O/S) implimentation.
Especially not before you've seen it :-)

I have a 100% Plug-and-Play computer system, and it took 3 reboots
to configure each device in Windows95! THIS IS UNACCEPTABLE! Heck,
DOS only took 2 reboots to configure everything! This is also
unacceptable though!

Now you know you can trust me - I *HATE* all the Pnp implimentations
I've used too! :-)

The PnP project encompasses:
- Intel/Microsoft Plug-and-Play
- PnP-ISA cards
- PnP-BIOS
- _maybe_ even the serial device 'COM' support
(for devices connected by a serial cable)
- PCI
- PCMCIA (hot swappable and docking stations even!)
- EISA
- Legacy (this is the trouble maker - DIE! DIE! DIE! )
- Intel ESCD (solves legacy problem most of the time)
If you hate all these, then I suggest you buy a Mac :^)

Select 'CONFIG_PNP=N' and you get a Linux kernel with no support
for any of the above, but you get a much *IMPROVED* resource
management system (replaces kernel/dma.c kernel/ioport.c and
touches other stuff) that takes up about the same space, and
doesn't have any races in it (unlike the current code).

Note: The current patch at the web site has the resource
managemnt code ready-to-run. It has many features driver
writers have been asking for - like address space management.

If you hate progress too, then look at CP/M :^)

-- Andrew E. Mileski --
--------------------------------------------------------------
mailto:dmtech@magi.com http://www.redhat.com/~aem/
"The best programmers are lazy", so I'm told.
I haven't gotten around to seeing if it is true or not though.