Re: Invalid output to process

[Milan Zamazal]

>>>>> "RMS" == Richard M Stallman <address@hidden> writes:

    >>     Why I can't just send you the content of `foo' variable?
    >> Because when I put its value into a buffer and then read it
    >> from the buffer, the read value is equalp to the original `foo'
    >> value, but the bug can't be reproduced with the new value.

    RMS> I agree that is perplexing, but that is no reason not to show
    RMS> us the value.  The information may be important--and I don't
    RMS> have time to try to figure out what the value is by reading the
    RMS> code.

    RMS> Would you please show us the printed form of that value?

Here is the output of (print foo):

output-print
Description: Binary data

    RMS> Have you reported a GDB bug?

Yes.

    RMS> Meanwhile, what exactly were the GDB commands that you used?
    RMS> Perhaps there is some other way to "debug the send_process C
    RMS> function" using different commands that won't encounter this
    RMS> bug.

The gdb commands I use initially are:

  break send_process
  run -q

Then I use `continue' repeatedly to reach the place of the bug
occurrence.  This works fine but after that gdb can crash as soon as
after a single `step' command or a few `next' commands.  But the crash
is irregular, so if you'd like to get some particular information, I may
be able to retrieve it.

    RMS> I suspect that this has something to do with character codings.
    RMS> I think that your output string is being encoded in some way.
    RMS> Please try binding coding-system-for-write to `no-conversion'
    RMS> around the calls to process-send-string and see if that fixes
    RMS> things.

It doesn't.  In both cases the input that server receives after applying
`M-x eval-current-buffer' in bug2.el is the same:

output-invalid
Description: Binary data

But if I use

  (set-process-coding-system process 'no-conversion 'no-conversion)

instead of

  (set-process-coding-system process 'raw-text-unix 'raw-text-unix)

after opening the process in bug2.el, the input is correct:

output-correct
Description: Binary data