[emacs-berlin] Literate dev-ops
andreas.roehler at easy-emacs.de
Wed May 11 16:55:44 UTC 2016
On 11.05.2016 14:55, Arne Brasseur wrote:
> On 11 May 2016 at 14:35, Rasmus <rasmus at gmx.us <mailto:rasmus at gmx.us>>
> Thanks for the link. Tramp is pretty cool.
> > #+NAME: build-new-uberjar
> > #+BEGIN_SRC sh :var VERSION=current-version :dir /
> > ssh:root at lambdaisland.com
> <mailto:ssh%3Aroot at lambdaisland.com>|sudo:web at lambdaisland.com:/var/web/app
> > scalar
> Aside: ob-sh is now ob-shell.
> Personally, those long lines babel lines are too complex for me,
> so I like
> to use the header keywords.
> #+header: :var version="1"
> #+header: :dir "/tmp"
> #+header: :results table
> #+begin_src shell
> echo $version, $(pwd)
> | 1 | /tmp |
interesting video, thanks Arne. Nonetheless have to shed some water into
As we've seen it's possible - but is it really easier then using a plain
Also: it isolates your setup and configuration from that of your
colleagues - in case its author is absent, it will be hard to take over
for the others in an enterprise.
And some more basic concern: the underlying literate programming
paradigm seems pointless WRT examples given in video. Why comments in
natural languages should be preferable over code? Code has the purpose
to be less ambiguous, it's a kind of math. While natural language gives
way to interpretation, i.e. misunderstanding of all kind. Maybe literate
programming was of interest in earlier times in a world of low level
coding. Higher languages are adapted at human mind, let's use them.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the emacs-berlin