On Thu, Oct 22, 2015 at 3:13 PM, Wolfgang Mauerer
<wolfgang.mauerer@xxxxxxxxxxxxxxxxx> wrote:
Am 22/10/2015 um 14:11 schrieb Mitchell Joblin:
On Thu, Oct 22, 2015 at 2:56 PM, Wolfgang Mauerer
<wolfgang.mauerer@xxxxxxxxxxxxxxxxx> wrote:
[trimmed CC list since this does not concern DB design any more]
Am 22/10/2015 um 13:51 schrieb Mitchell Joblin:
On Thu, Oct 22, 2015 at 2:50 PM, Mitchell Joblin <joblin.m@xxxxxxxxx>
wrote:
On Thu, Oct 22, 2015 at 12:42 PM, Wolfgang Mauerer <wm@xxxxxxxxxxxxxxxx>
wrote:
do you get reasonable scatterplots from the initiate-response plug in?
This warning simply says that for a given thread, there are no
responses. We should maybe get rid of this message completely.
I don't think that's right. I think this statement is executed when
all threads have no responses. The condition for executing the
statement is "is.null(dat.replies)" and dat.replies contains all
replies for all email threads, not just a single one.
you're right, I should have looked closer. Still, does the scatter plot
contain any useful information? For qemu, for instance, I get reasonable
results. Which list are you working with?
I forgot to add that I checked for scatter plots in the web frontend
and nothing is available there.
A was referring to plots like the following:
p
http://im-vm-010.hs-regensburg.de:8081/apps/details/?projectid=85&widget=widget.ir.details.receiving,widget.ir.details.writing&topic=communication
Yes, this is the widget I checked.
okay. I was not sure if you're referring to the widget or the content
when you said nothing was there.
Since the only difference between our setup seems to using a
release-base versus a sliding window-based analsis, could you
run your analysis with explicit releases?
Thanks, Wolfgang
Perhaps you need to add the widget manually using the widget selector
in view "communication". I'm working with git HEAD on a vagrant
deployment.
I have tried this for ivy and qemu (last year of development). Could
you please share your mbox file? Also, are the results you have in
your database from a relatively recent run with up-to-date R packages?
I am also running the analysis after using a date based proximity
analysis. Perhaps there is some unaccounted dependency there? The only
coupling I see between the VCS and mailing list analysis is by using
the revision dates but that shouldn't be any different between the VCS
analysis methods.
ivy worked fine for me with proximity analysis based on explicit
releases. I'll send the mbox file in a private message.
Great.
Thanks,
Mitchell
Thanks, Wolfgang
--Mitchell
Thanks, Wolfgang
