Hi Björn,

On Thu, Apr 7, 2016 at 4:09 AM, Björn Dahlgren <[email protected]> wrote:
>
>
> On Sunday, 30 August 2015 23:00:43 UTC+2, Aaron Meurer wrote:
>>
>> That sounds like a good idea. Having them in a separate repo would make it
>> less of a big deal to include inline plots in the notebooks.
>
>
> Yes, I have strong feelings against checking in any inline plots (or
> generated binary data in general) into the SymPy repository.
> The issue recently surfaced here: https://github.com/sympy/sympy/pull/10869
>
> Regarding doctesting the actual notebooks: I have been doing this in my
> projects for a while (rendering notebooks as part of CI process, thus never
> checking in any output data). See e.g.
> http://hera.physchem.kth.se/~chempy/branches/master/examples/ for an
> example.
> The rendering (and testing) is achieved by using nbconvert:
>
> ipython2 nbconvert --to=html --debug --ExecutePreprocessor.enabled=True
> *.ipynb
>
> In case someone wants to try to set this up (the CI servers would need to
> push artifacts somewhere on success).

I converged to the same workflow. I write a simulation code in Fortran
(check into git), which generates a log file (not checked in). Then I
have an jupyter notebook that parses the log file and generates
graphs. I check in the notebook, but without output cells. Then it's
small, the "git diff" is very readable (say if I improve a plot in a
new commit) etc.

Finally, and I don't have it setup yet, the goal is for the CI to run
the test (Fortran code) on some very small problem (for example, if I
use FFT and I need 256^3 and 20,000 time steps to get fully converged
results, I would run it with 16^3 PW and 20 time steps, so that it
runs in few seconds), and then I test the notebook that it runs using
nbconvert, exactly as you posted, to check that the analysis still
runs.

That way things should be 100% robust, and I can regenerate the graphs
any time I want, yet no big files or binary blobs are checked into a
git repository.

Ondrej

-- 
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sympy.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sympy/CADDwiVCGTNz-aEsOBkvep9Sz7AXsN8AKeRJ%3Dx9H2B_uDU6dU6g%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to