S.J.Eglen at damtp.cam.ac.uk
Sun Dec 11 14:36:27 CET 2011
> I can see how all this stuff will pay back in the long run, but for
> the time being it's a huge paradigm shift. Integrating swankr into ESS
> would mean shipping slime with ESS, with an advantage of virtually
> seamless transition from users' perspective, and disadvantage of
> requiring a huge investment from ESS team to refactor the existing
> interface and learn the internals of swankr and slime. If a
> stand-alone R-mode to be developed, that would mean a split in user
> base, and obviously users will have to learn slime interface. Moreover
> developers of external addons will have to decide what interface to
> support and thus a split in developer base - always a bad thing. Also
> I am not sure how welcome R is for slime developers. Presumably the
> nature of R will require some specific tailoring in the slime itself,
> which means some sort of cooperation will be required.
While I think these are all valid concerns, at this stage I for one am
interested in seeing what there is to gain from switching from comint to
sockets for communicating with R. The comint code in ESS works, but is
definitely creaking, and has bugs that we all have had to work around.
Another thing (that Martin has pointed out recently) is that ESS has
poor support for much of the recent changes to R's help system. If
using SLIME works out to be a success, in the first instance I'd vote
for asking users to install SLIME separately (rather than bundling it
with ESS), e.g. via the package manager coming in Emacs 24.
The way Christophe has it set up it sounds clean, to allow us to explore
using SLIME as an alternative to comint, whilst keeping ESS running for
other aspects (e.g. code/doc editing).
Christophe: is there anything on the ESS side that could be done to make
your life easier, without harming ESS users who still use comint?
More information about the ESS-help