[petsc-dev] PETSc statistics

Jed Brown jed at jedbrown.org
Thu Jun 8 23:33:32 CDT 2017

Sean Farley <sean at farley.io> writes:

> Jed Brown <jed at jedbrown.org> writes:
>> Sean Farley <sean at farley.io> writes:
>>> Jed Brown <jed at jedbrown.org> writes:
>>>> Sean Farley <sean at farley.io> writes:
>>>>> Sean Farley <sean at farley.io> writes:
>>>>>> Andrew McRae <A.T.T.McRae at bath.ac.uk> writes:
>>>>>>> But we clone from our fork (formerly Bitbucket firedrake/petsc, now Github
>>>>>>> firedrakeproject/petsc), not from Bitbucket petsc/petsc.
>>>>>> Yeah, the stats won't be perfect, of course. Hopefully, useful enough
>>>>>> for some general idea of the numbers.
>>>>> Also, there are harder questions to answer: does a pull count? If so,
>>>>> what about a no-op pull? A pull of one commit? etc.
>>>> It would be fascinating to know how often people issue pulls.  For
>>>> example, we really have no idea how many people track 'master' or
>>>> 'next', but daily pulls, particularly those not at exactly the same time
>>>> each day (e.g., cron/nightly testing versus issued by a human) would
>>>> help give a picture.
>>> Fair enough. Attached is a new PDF with that data. As you mention,
>>> though, there's no good way to tell a pull is from a bot. 
>> Unique IP?  Doesn't help for on-demand CI systems, but would be relevant
>> to institutional machines.
> Sure, I can do unique ips if you want. Previously, you wanted "how often
> people issue pulls". If I do two pulls in one hour, how would you like
> that counted? Two pulls in one minute?

There are lots of ways to present the data and I don't know what would
be most informative.  Would it be possible to download some quasi-raw
representation of the log files?

> Did you also want forks?

Maybe, but probably not too substantial for PETSc (since we don't use a
particularly fork-heavy workflow).

> If you can come up with database-like queries / well-defined metrics /
> questions, I can do my best to get that for you guys. Ping me on chat /
> IRC (#bitbucket) if you want to talk real-time without the delay.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20170608/92d56ddc/attachment.sig>

More information about the petsc-dev mailing list