[PATCH] pnp.ids

[Richard Hughes]

On Mon, 07 Feb 2005 18:30:05 +0100, Pozsár Balázs wrote:
> NO NO NO NO NO!!!

Err, can I stand nearer the fence and say "probably not a good idea.."?
:-)
 
> Please do not do this. I really think this is a bad idea, duplicating a
> constantly changing database. Why not just read those files on startup
> into a binary tree? Would give the same performance, without just
> another duplication/"fork".

My gut reaction is that regenerating a .h file is a *bad* thing to do,
but is this file large? I guess it'll be a little over 10k memory then
whats the problem with caching it? Or even letting the kernel do the
caching? Even my hard-disk has a 8Meg cache... :-)

If the data is indeed shared between hal and pcmcia-cs, then maybe we
should come to some agreement between the both of us.

I think personally, that pnp.ids should live in /usr/share/hwdata and be
installed by hwdata, just like pci.ids and usb.ids - but that does put a
dependency on hwdata that maybe David is keen to avoid. We can't have one
source of information in pcmcia-cs, some on the website, some in the
kernel, and some in a header file for HAL. If we made pnp.ids part of a
common package, say hwdata then a typical user could bugzilla a report
against hwdata, and an update sent for just hwdata. Then ALL the tools
could benefit.

Just my point of view. Note there's nothing technically wrong having HAL
or hwdata own the /usr/share/hwdata/pnp.ids file - pcmcia-cs already
depends on hwdata >= 0.60 so would be trivial to change.

Richard.

-- 

http://www.hughsie.com/PUBLIC-KEY


_______________________________________________
hal mailing list
hal at lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/hal