Should launchpad hwdb fingerprints be unique?

Asked by Tommy Trussell on 2012-08-05

I signed up to test some experimental kernels at packages.qa.ubuntu.com and I noted that others were submitting system fingerprints of some sort. SO I just ran checkbox on two different laptops I own, very different ages, different screen sizes, different processors, different graphics chipsets, different wireless chipsets, slightly different keyboard layouts, everything is different on the two laptops EXCEPT they are both ACER brand units.

I was surprised when checkbox didn't ask me to test lots of things, such as suspend on either unit. (Ironically the first laptop, a newer netbook, suspends fine, most of the time, whereas the old one suffers from a resume bug in its Radeon graphics chip that makes it unusable for suspend at all.) Checkbox chose several tests not appropriate for the hardware, and left out several that it should have included.

HOWEVER, when I looked at my hwdb submissions for these two different laptops, they generated the SAME fingerprint 2954e74ba17fb0e37fc942cd1d9fab4e See https://launchpad.net/+hwdb/+fingerprint/2954e74ba17fb0e37fc942cd1d9fab4e -- as of tonight there are 6609 tests with this fingerprint, and I wonder if they are being lumped together inappropriately.

Should I expect the hardware "fingerprint" to be unique to unique hardware, or are the fingerprints being aggregated in a way that is not immediately obvious?

Question information

Language:
English Edit question
Status:
Solved
For:
Checkbox Edit question
Assignee:
No assignee Edit question
Solved by:
Marc Tardif
Solved:
2012-08-16
Last query:
2012-08-16
Last reply:
2012-08-16

I'm happy to create a bug report with the lshw output and other info as needed.

Best Marc Tardif (cr3) said : #2

The fingerprint should be unique to the hardware model, not your instance of that model. If this is not the case, then this question should probably be converted to a bug.

Nicholas Skaggs (nskaggs) said : #3

Tommy, I can't answer the checkbox side of the question. However, for submitting a result on the tracker, you can simply issue an lshw and paste the output onto something like paste.ubuntu.com and then use the URL in the field instead. You can of course remove personal serial's or mac addresses's before pasting if you wish.

Hi -- for the bug report against checkbox I'm using lshw -sanitize > lshw.log and attaching the logfile to the bug. It looks like others were doing the same sort of thing and dropping the file in UbuntuOne or a dropbox folder.

Thanks for the suggestion!

I'm going to mark this question as solved-- it looks like the hwdb fingerprints SHOULD be unique, so there's a bug somewhere. Thank you!

Thanks Marc Tardif, that solved my question.

For anyone bumping into this message, be sure to read the related bug -- the fingerprint is rarely used and is not generated in a way that would make it unique. Different hardware should be identified a different way. (I'm still not sure why I was having trouble getting checkbox to check its hardware appropriately but apparently that's a different issue.)

Marc Tardif (cr3) said : #8

Please don't hesitate to report a bug about getting checkbox to check hardware properly. If it's doing something that you're not expecting or, inversely, not doing something that you're expecting, please let us know. Thanks!