[PATCH] pnp.ids

[Pozsár Balázs]

On Mon, Feb 07, 2005 at 04:45:14PM -0500, David Zeuthen wrote:
> Sjoerd Simons wrote:
> >What's wrong with using files with fdi-like format for this.. Hal already 
> >has a
> >parser for those and it's so much more flexible..
> >
> 
> Exactly - and one day we can download .fdi files on demand which is 
> helluva lot more flexible that getting some /usr/share/hwdata/pci.ids 
> file updated.
> 
> Woow - Sjoerds view is even a bit more extreme than mine!

I don't like this download-on-demand thingy... why not simply
"just work", or, if not, update the system packages and then just work?

Packages already have a good and reliable update mechanism. Why 
introduce another system which does not give _any_ new capabilites over 
what we have now?


> >But then again, i've got a dislike for data build into binaries :)
> 
> These data represent facts; they only *change* if there is a bug. 
> Consider it an optimization - parsing XML is much slower than using an 
> O(log n) algorithm to look a PCI id table. And there are plenty PCI and 
> USB id's :-).

Exactly how slow is the lookup currently? I can't believe it is 
bottleneck.
But if it is, any .ids (or .fdi) file have only be parsed/read only 
once, on hald startup (maybe on sighup too), and takes only a fraction 
of a second. I wish this would be the worst sw bloat on my system...

The pci.ids and usb.ids do change over time, mostly by covering new 
entries, but there are usually fixes to the older ones too. If the 
hwdata package ships these files, moreover an uptodate version of them, 
why duplicate? I really hate duplication, and this would be tipical ugly 
one. Have ONE place in a system where you have the updatetodate and 
most correct information about a particular thing (ie pci/usb ids). And 
we already have that place, and pnp ids are exactly the same category.


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