bug#40323: 28.0.50; error in process filter: Invalid search bound (wrong side of point)

[Jacob Lagares Pozo]

Here is an example output from Slack, even though weirdly enough nothing seemed to trigger the error:

    ======================================================================<br
      />
      1 -&gt; (comint-output-filter #&lt;process Shell&gt;
      &quot;[05/12/20, 13:06:14:636] info: [DESKTOP-SIDE-EFFECT]
      Update from desktop for keys
      &nbsp;[\&quot;settings\&quot;]&nbsp;<br />
      &quot;)(:comint-pmark nil)<br />
      | 2 -&gt; (set-marker #&lt;marker at 37370 in *Async Shell
      Command*&gt; 37469)(:comint-pmark (#&lt;marker at 37370 in
      *Async Shell Command*&gt; . #&lt;marker at 37469 in *Async
      Shell Command*&gt;))<br />
      | 2 &lt;- set-marker: #&lt;marker at 37469 in *Async Shell
      Command*&gt;(:comint-pmark (#&lt;marker at 37469 in *Async
      Shell Command*&gt; . #&lt;marker at 37469 in *Async Shell
      Command*&gt;))<br />
      | 2 -&gt; (set-marker #&lt;marker at 37469 in *Async Shell
      Command*&gt; 37566)(:comint-pmark (#&lt;marker at 37469 in
      *Async Shell Command*&gt; . #&lt;marker at 37469 in *Async
      Shell Command*&gt;))<br />
      | 2 &lt;- set-marker: #&lt;marker at 37566 in *Async Shell
      Command*&gt;(:comint-pmark (#&lt;marker at 37469 in *Async
      Shell Command*&gt; . #&lt;marker at 37566 in *Async Shell
      Command*&gt;))<br />
      | 2 -&gt; (ansi-color-process-output &quot;[05/12/20,
      13:06:14:636] info: [DESKTOP-SIDE-EFFECT] Update from desktop for
      keys &nbsp;[\&quot;settings\&quot;]&nbsp;<br
      />
      &quot;)(:comint-pmark (#&lt;marker at 37469 in *Async
      Shell Command*&gt; . #&lt;marker at 37566 in *Async Shell
      Command*&gt;))<br />
      | 2 &lt;- ansi-color-process-output: nil(:comint-pmark
      (#&lt;marker at 37469 in *Async Shell Command*&gt; .
      #&lt;marker at 37566 in *Async Shell Command*&gt;))<br
      />
      | 2 -&gt; (set-marker #&lt;marker (moves after insertion)
      at 37566 in *Async Shell Command*&gt; 37566)(:comint-pmark
      (#&lt;marker at 37469 in *Async Shell Command*&gt; .
      #&lt;marker at 37566 in *Async Shell Command*&gt;))<br
      />
      | 2 &lt;- set-marker: #&lt;marker (moves after insertion)
      at 37566 in *Async Shell Command*&gt;(:comint-pmark
      (#&lt;marker at 37469 in *Async Shell Command*&gt; .
      #&lt;marker at 37566 in *Async Shell Command*&gt;))<br
      />
      1 &lt;- comint-output-filter: #&lt;marker (moves after
      insertion) at 37566 in *Async Shell Command*&gt;(:comint-pmark
      nil)<br />

Because I am on the testing branch of Debian, is there any chance Emacs got updated and this bug fixed?

    On 05/05/2020 19:33, Noam Postavsky wrote:
> Jacob Lagares Pozo <jlagarespo@iebesalu.cat> writes:
>
> > If I run it without your patches, it works surprisingly just fine (I
> > noticed the original errors pop up most commonly on Slack I guess
> > because it prints a lot more?), whereas if I evaluate said patches,
> > this is the output of the trace buffer:
> >       
> I don't see anything unexpected in the trace either.
>
> > I am not sure what does this mean, perhaps it is some special
> > character Slack uses for logging that messes with those markers, I
> > don't know. Maybe I could try printing all of the ASCII characters
> > sequentially and see what happens.
> >       
> Well, ideally we would want a trace from something that does trigger the
> error.
>
> > Regardless, I'm still not entirely sure what your code is doing anyway.
> >       
> It's basically just printing out the values of the markers, so we might
> hopefully notice when they get changed in an unexpected way.
>