One would think!  I tracked the issue to the `sql-redirect-one'
function, which is responsible for re-populating the buffer.
Unfortunately I wasn't able to fix the problem for myself.

I tested including an (inhibit-read-only t) in the opening let form,
but still no joy.  Seems the call to
`comint-redirect-send-command-to-process' redirects output to the
target buffer then calls `process-send-string' (shocking, I know) and
that *latter* call gives the same error.

I don't know nearly enough about Emacs, but my current best guess is
that `process-send-string' just won't respect `inhibit-read-only'
because the former's a built-in C function.  Right now I'm trying to
figure out how best to manage the read-only status on the buffer (I'm
really new to the nuances of view-mode and toggling the read-only
flag).

Any pointers?  Right now I'm just jamming a (read-only-mode -1) and
(read-only-mode 1) at the top and bottom of `sql-redirect-one', but it
feels too gross to share.  I'd love to find out that I can get away
with a simple `inhibit-read-only' solution.



On Thu, May 30, 2013 at 9:35 AM, Stefan Monnier <monnier@iro.umontreal.ca>wrote:

> > schema and table elements as I learn the database layout).  If instead
> > of quitting out of view-mode I just switch back to my sql-mode buffer,
> > the second call to the sql-list-foo command gives a "Buffer is
> > read-only" error.
>
> Sounds like a bug.  Either it shouldn't use view-mode, or (more likely)
> it should let-bind inhibit-read-only around the code that (re)populates
> the buffer.
>
>
>         Stefan
>