you are right, we have also discussed about this issue inside the mailing list where we have thought that a good solution was to use the vendirid:productid code together with the revision number.

Anyway, I think that, since bcm4311 is the only case in which we have faced this problem, perhaps it could have been an error and that case could be the unique exception. I was waiting to see if we discover other similar cases. For now the fact that the revision 1 works is stated inside the description entry.

So I think that, before adding a new option in the select, we could wait and see if other similar cases arise. What do you think?

Perhaps we could add the revision entry for all the devices even if we decide not to use it to check the chipset (together with the vendorid:productid code). This modification has to be done on both the server (h-source) and the client (h-client), so it requires a bit much time :)

The thing about revisions is that some of them are as usual as: rev 01, rev 02, rev 03, etc, but some other aren't quite standard.

Like:

IMHO, we are reaching the point that we can't try to update the whole db with new info from the devices already in it.

I think the revision number is a hexadecimal number, perhaps we could allow all the combinations among 00 and ff. But I have to check if this is true.

IMHO, we are reaching the point that we can't try to update the whole db with new info from the devices already in it.

Yes it's true, and I hope we have chosen the right way to catalogue hardware :)