[Insight-developers] Re: Dashboard

David Cole david.cole at kitware.com
Wed Jan 31 12:12:03 EST 2007


So, the lesson for us all is:
"If you're going to take a shortcut to save some of your own time, make sure
you do it when there's a continuous dashboard that's got your back.
Otherwise, it might make the project look bad and make other people spend
their time investigating problems caused by my shortcut."

:-)

Any other comments on the 3pm guideline?

Thanks,
David Cole

On 1/31/07, Julien Jomier <julien.jomier at kitware.com> wrote:
>
> Dave,
>
> It did compile on my machine but using explicit instantiation and since
> it takes a significant amount of time to recompile ITK without explicit
> instantiation, I missed this one.
>
> 3pm sounds a reasonable time to hold checkins.
>
> Julien
>
> David Cole wrote:
> > On the other hand, after investigating the history a little bit more, I
> > have some questions and comments:
> >
> > Question:
> > Julien, did this compile on your machine at the time you committed these
> > changes? It looks to me like it compiles literally *nowhere* based on
> > the nightly results. If it did compile for you, then this is truly a
> > mystery... if not, then, tsk tsk -- always compile before committing.
> >
> > Comment:
> > 6:39 pm eastern time is too late to commit code changes to the ITK code
> > base. Most of the continuous dashboards stop building/submitting around
> > 6pm eastern (they typically start around 6am eastern and run for 12
> > hours...)
> >
> > Question for all of you ITK developers:
> > Is 3pm eastern time a reasonable guideline for "last allowed commit time
> > before the nightly start time"? That would give the continuous
> > dashboards plenty of time to do their thing and notify folks by email of
> > any problems with their commits. Then there would still be a chance to
> > commit a fix for a problem prior to the nightly runs.
> >
> > There is *no* reason why you can't hold off on a commit until the
> > following day. You can get all your changes ready to go and then, if
> > it's too late in the day, simply wait until the next morning to do the
> > commit. Or at least until after the nightly start time.
> >
> > I will add a page to the ITK Wiki that contains some "cvs commit
> > guidelines" for ITK developers. I did this for VTK a while back,
> > including the recommendation *not* to commit past noon eastern time, and
> > it's had more than 1500 hits in the last 3 months. And people are
> > largely following the guideline lately. (And it's just a guideline. It
> > won't be mechanically enforced -- just make sure you really have a good
> > reason to violate it if you do.)
> >
> > Comments? Does anybody strongly object to a "3pm eastern commit time"
> > guideline?
> >
> >
> > David Cole
> > Kitware, Inc.
> >
> >
> > On 1/31/07, *David Cole * <david.cole at kitware.com
> > <mailto:david.cole at kitware.com>> wrote:
> >
> >     No problem... It wouldn't be fun or challenging if somebody didn't
> >     totally make it red *once in a while*... We'll make this your one
> >     "grace". Next time, I'll have to get rough... ;)
> >
> >
> >     On 1/31/07, *Julien Jomier* <julien.jomier at kitware.com
> >     <mailto:julien.jomier at kitware.com>> wrote:
> >
> >         Dave,
> >
> >         I just fixed the problem with the dashboard.
> >         The continuous builds should be green soon.
> >
> >         I'm adding more template instantiations (I'm almost there) so I
> >         might
> >         break the dashboard, but I'll try my best to keep it green.
> >
> >         Sorry about that,
> >
> >         Julien
> >
> >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.itk.org/mailman/private/insight-developers/attachments/20070131/60e47258/attachment.html


More information about the Insight-developers mailing list