Re: missing data in relaxation dispersion -- May 02, 2016 - 17:37

On 29 April 2016 at 11:58, Petr Padrta <padrta@xxxxxxxxxxxx> wrote:
> On Wed, Apr 27, 2016 at 11:27:09AM +0200, Edward d'Auvergne wrote:
> > On 22 April 2016 at 20:43, Petr Padrta <padrta@xxxxxxxxxxxx> wrote:
> > > Hi Edward,
> > >
> > > OK,  I finally managed to create a bug report on gna with attached 
> > > example.
> >
> > Hi Petr,
> >
> > Thank you for that.  For reference, the report is at
> > https://gna.org/bugs/?24601 .  I have taken your data and script and
> > created the following relax system test:
> >
> > $ relax -s Relax_disp.test_bug_24601_r2eff_missing_data
> >
> > I used this to catch the bug you saw and to fix all the issues.  Note
> > that it took a bit longer than normal as the first error you saw was
> > only one of 3.  I also had to update the Monte Carlo simulation error
> > analysis code, and the Grace plotting code.  I have committed the
> > changes:
> >
> >     http://article.gmane.org/gmane.science.nmr.relax.scm/25955
> >
> > When using this, please carefully check that all is working as
> > expected.  The system test now passes, however there might be other
> > strangeness (for example, check your Grace graphs for incorrect data
> > labels).  For any other issues you see, if you could create a separate
> > bug report for each, that would be appreciated.  As before, if you
> > could include truncated data and a script to replicate the issue, that
> > would be much appreciated as I could then create the system test to
> > help fix the problem.  The more bugs reported - from critical to
> > superficial - the smoother we can make the analysis process!  Even
> > confusing text messages by relax deserve a report.
> >
> > For obtaining the code, if you need this soon, I would suggest
> > directly obtaining it from the source code repository:
> >
> >     http://www.nmr-relax.com/download.html#Source_code_repository
> >
> > The reason is because there are a few newly introduced bugs that I
> > would like solved before we release relax 4.0.2.  So it might take a
> > few weeks before the new release is out.
> >
> > Regards,
> >
> > Edward
>
> Hi Edward,
>
> Thanks! I fetched last SVN version (r28204) and so far it seems that your 
> new
> dispersion code works as expected. The (rhetorical) question is what to do 
> with
> interpolated dispersion curves from missing B0 fields - they have the right
> shape but are shifted along R2eff-axis as their r20 is unknown. Hmm, I can
> always delete them from the grace plots. Or maybe I'll do another bug report
> after some more testing.

Hi Petr,

A new bug report for such things would be much appreciated.  I can
then list the bugs one after the other and their details in the
release notes.  It also makes it incredibly easy for a developer to
fix the bug if truncated, minimal data sets and a script and/or
instructions are attached to the bug report.  That way a system test
can be quickly constructed to reproducibly capture the bug.  In most
cases where a system test can be set up, the fix then only requires
5-10 minutes to resolve.

Cheers,

Edward