[ESS] Prevent ESS From Writing .Rhistory

[Stephen Berman]

On Thu, 08 Sep 2016 22:06:34 +0200 Stephen Berman <stephen.berman using gmx.net> wrote:

> On Thu, 08 Sep 2016 19:12:51 +0100 Stephen Eglen <sje30 using cam.ac.uk> wrote:
>
>>> For me it's the same as for Rich Heiberger: dots when entering the
>>> password in the minibuffer but then it gets printed in cleartext after
>>> the "Password: " prompt in the *R* buffer.  This is with R version 3.3.1
>>> Patched (2016-09-02 r71210) Platform: x86_64-suse-linux-gnu (64-bit),
>>> ESS 16.04 built today from the Git repository (commit 14e80d60) and with
>>> both a recent Emacs build from the Git master branch as well as Emacs
>>
>> hi Steve,
>>
>> interesting - this is the first report on linux that I'm aware of.
>>
>> Can you try something like the following:
>>
>> M-x shell
>>
>> and then in that *shell* buffer try to ssh somewhere that will ask you
>> for a password.
>>
>> Does the *shell* also echo the password?
>
> No, the password is not printed in the *shell* buffer.  If you can tell
> me what ESS function to instrument, I can try edebugging.

I found the cause of the password echoing: the "Process Echoes" menu
item in the iESS->Process menu is checked.  This is also the case when I
load ESS in Emacs started with -Q.  Both with -Q and with my
initializations, comint-process-echoes is nil (the default), and the
above menu item reverses this value, activating echoing.  So the obvious
question is, why isn't this the case for those ESS users (apparently
almost everyone) who don't get the password echoing?  For these users,
it seems comint-process-echoes is (changed to) non-nil when ESS is
loaded; why -- or why doesn't that happen for me?

Steve Berman