Matthew Landis <lan...@isciences.com> writes: > <cberry <at> tajo.ucsd.edu> writes: > >> >> Eric Schulte <eric.schulte <at> gmx.com> writes: >> >> >>> Does this do what you want? > >> > >> > Have you looked at the :cache header argument [1], from my understanding >> > of your use case it should be exactly what you are after. >> > >> >> Its a step in the right direction. >> >> It seems I have to set :cache yes on every block I use before I invoke >> it. My attempt to use a buffer-wide PROPERTY setting for cache did not >> pan out. >> > > I'd like to put in a vote for the kind of functionality that cberry is > describing. I have a very similar situation - a large org file that uses R > to > do a lot of time consuming data manipulation and model fitting, resulting in > statistical tables and graphs. I run a lot of the code blocks as I'm writing > it, resulting in :results in the org file. > > In the end, I'd like to export the org file to html or ODT, but I'd like to > be > able to choose buffer-wide whether to rerun all of the code blocks or just > use > the results that are already in the buffer. I tried setting #+PROPERTY: eval > no > at the top of the buffer in the hopes that on export, it would ignore all my > code blocks and just incorporate the :results, but this was ignored and my > code > blocks were rerun. > > The cache argument only partially deals with the problem, as this example > illustrates: > > #+begin_src R :session :cache yes > x <- rnorm(100) > #+end_src > #+begin_src R :session :results graphics :exports results :file hist.png > :cache > yes > hist(x) > #+end_src > > Now after the first export, I change code block 2, but not code block 1. If > I > understand how cache works correctly, code block 2 will be rerun, but it will > fail because code block 1 is not rerun, so x doesn't exist in the R session. > > For this reason, I'd prefer to be able to decide whether to re-run on a file- > wide basis. > > Many thanks to all of you who have created such an amazing system. > > M >
Matthew, I think that you're wrongly expecting babel's cache header argument to behave like the argument of the same name in Sweave code chunks. Babel will cache, in your case, the value of your code block evaluation and there is none in your first code block, therefore nothing gets cached by babel, try that instead: #+name: my-random-vector #+begin_src R :session :cache yes rnorm(100) #+end_src #+headers: :var x=my-random-vector #+headers: :results graphics :exports results :file hist.png #+begin_src R :session :cache yes hist(x) #+end_src Does it work better? In that case you don't even need a session. Christophe -- Président, Nicolas Sarkozy représente une sorte de triomphe bouffon de l'égalitarisme français ; pour la première fois de notre histoire, nous avons un chef de l'État qui se comporte comme s'il ne valait pas mieux que les citoyens. C'est en réalité toujours le cas, mais cette vérité doit être cachée pour que les institutions et le système social tournent de façon, si ce n'est harmonieuse, du moins raisonnable. E. Todd, Après la démocratie. -- Christophe Pouzat MAP5 - Mathématiques Appliquées à Paris 5 CNRS UMR 8145 45, rue des Saints-Pères 75006 PARIS France tel: +33142863828 mobile: +33662941034 web: http://www.biomedicale.univ-paris5.fr/physcerv/C_Pouzat.html
