Hi Stefan,

They are indeed intended to be seen by the user. They hold important
communications between server and client, and I ask users to post their
contents when making Eglot bug reports.

Well I suppose the stdout one less useful than stderr.

And this is more of an LSP requirement than an jsonrpc one. I could
theoretically make Eglot tell. Jsonrpc to promote the buffer's visibility,
but that seems like overkill right now.

João



On Fri, Dec 13, 2019, 19:28 Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> Hi João,
>
> >     Disable undo in the process buffers of a JSONRPC connection
> > +      (buffer-disable-undo)
> >        (set-marker (process-mark proc) (point-min))
> >        (let ((inhibit-read-only t)) (erase-buffer) (read-only-mode t)
> proc))
> > +    (with-current-buffer stderr
> > +      (buffer-disable-undo))
>
> Assuming those buffers are only used internally and are not intended to
> be displayed to the end-user (other than for debugging purposes), then
> using buffer names with a leading space should take care of this
> (undo is disabled by default in those buffers).
>
> If you're already using a leading space in those buffers's names, then
> I wonder how/why their undo was enabled in the first place.
>
>
>         Stefan
>
>