bug#65156: 29.1; Reading from pipe with --insert or insert-file-contents no longer supported

[Eli Zaretskii]

> Date: Wed, 9 Aug 2023 13:57:36 -0700
> Cc: larsi@gnus.org, 65156@debbugs.gnu.org
> From: Paul Eggert <eggert@cs.ucla.edu>
> 
> On 2023-08-08 12:27, Eli Zaretskii wrote:
> > Paul, can there be a regular file that is not seekable?
> 
> No.

Thanks.  This is no longer an issue, though.

> The current code on master is a bit of a mess. There's nothing wrong 
> with using a positive BEG on a seekable and non-regular file, for 
> example; the old doc string was wrong and now the code has been changed 
> to match the doc string unfortunately. Nor is there anything wrong if 
> BEG is 0 on a non-seekable file (though this latter issue is 
> longstanding and so I guess nobody cares).

These are minor issues that should be easy to clean up, I think?  If
so, now is a good time for doing that on master.

> There are surely some gotchas involving the REPLACE arg of 
> insert-file-contents too, but it's hard to tell because the doc string 
> seems to be corrupted for the case where REPLACE is 'if-regular' and I 
> don't know what it's trying to say.

I guess you are alluding to this part:

  If REPLACE is the symbol ‘if-regular’, then eschew preserving marker
  positions or the undo list if REPLACE is nil if FILENAME is not a
  regular file.  Otherwise, signal an error if REPLACE is non-nil and
  FILENAME is not a regular file.

Which part(s) of this are unclear?