Comments regarding pci.ids_0.0~2019.06.30-1_amd64.changes

[Martin Pitt]

Hello Aurelien,

sorry for the latest reply!

Aurelien Jarno [2019-09-06 20:04 +0200]:
> > > 	Is pci.ids in its preferred form for modification?  The upstream git
> > > 	history suggests that this file is being automatically generated from
> > > 	the database at <https://pci-ids.ucw.cz/>.

The database itself does not seem to be accessible, though. Even that site
presents the textual pci.ids file as *the* reference artifact to download.

> > > Wouldn't someone wanting
> > > 	to edit pci.ids want to run want to modify the database and run the
> > > 	same code upstream runs to generate the new pci.ids file?

> > The source here is IMO a bit fuzzy, given that this is factual data.
> > There's a website, the purpose of which is to make it easy for the
> > current maintainers to incorporate proposed entries from third parties.

Agreed -- in that sense, the real *source* is the bureaucracy involved that
hands out new vendor IDs (external) and the vendor-internal process to assign
new product IDs. So if you really want to make changes there, you need to
engage with the vendor, and given how these are just a convention and factual
data, this makes no sense. The "normal" case would be for someone to spot a
mistake in pci.ids, in which case it should certainly be reported to that
database.

> > So, I guess very strictly speaking this is not the "source"

"source" is rather philosophical for this case; but from the perspective of
some package maintainer wanting to fix e. g. a typo in a product ID, this text
file very much *is* the preferred form of modification :)

> Adding the systemd maintainers in Cc: as both usb.ids and pci.ids are in
> the systemd source package.

Thanks for the notification. However, it seems we all agree that the current
status quo is fine?

Martin