Re: [patch] Fix python gdb.execute to not paginate

[Doug Evans <dje@google.com>]

On Thu, Aug 5, 2010 at 2:20 PM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> Hi,
>
> downstream Bug: https://bugzilla.redhat.com/show_bug.cgi?id=620930
> (gdb) python for i in range(100): a = gdb.execute('info registers', to_string=True)
> ---Type <return> to continue, or q <return> to quit---
>
> When at it I have merged it with making --batch more batch, as the output IMO
> should not depend on the momentarily terminal - wrapping and indenting the
> output on its width.
>
> The patch makes python gdb.execute running in the --batch mode.  I understand
> it may not be universally right but so far I believe it is.  What do you
> think?
>
> (--batch should be IMO somehow merged now with `set interactive-mode'.  That
> is not a part of this patch.)
>
>
> Regards,
> Jan
>
>
> gdb/
> 2010-08-05  Jan Kratochvil  <jan.kratochvil@redhat.com>
>
>        * defs.h (make_cleanup_restore_uinteger)
>        (make_cleanup_restore_page_info): New declarations.
>        * python/python.c: Include main.h.
>        (execute_gdb_command) <to_string>: Temporarily set BATCH_FLAG and call
>        init_page_info.
>        * utils.c (make_cleanup_restore_uinteger)
>        (init_page_info) <batch_flag>
>        (do_restore_page_info_cleanup, make_cleanup_restore_page_info): New.
>
> gdb/testsuite/
> 2010-08-05  Jan Kratochvil  <jan.kratochvil@redhat.com>
>
>        * gdb.python/python.exp (show height, set height 10)
>        (verify pagination beforehand, verify pagination beforehand: q)
>        (gdb.execute does not page, verify pagination afterwards)
>        (verify pagination afterwards: q): New.
>
> gdb/doc/
> 2010-08-05  Jan Kratochvil  <jan.kratochvil@redhat.com>
>
>        * gdb.texinfo (Mode Options) <-batch>
>        (Basic Python) <gdb.execute>: Describe setting width and height.

Yikes, nasty problem.

@@ -380,6 +381,13 @@ execute_gdb_command (PyObject *self, PyObject
*args, PyObject *kw)

      if (to_string)
       {
+         /* GDB_STDOUT should be better already restored during these
+            restoration callbacks.  */
+         make_cleanup_restore_page_info ();
+         make_cleanup_restore_integer (&batch_flag);
+         batch_flag = 1;
+         init_page_info ();
+
         str_file = mem_fileopen ();

         make_cleanup_restore_ui_file (&gdb_stdout);

IWBN to bury the implementation details.  I.e. move the setting of
batch_flag into utils.c.  Plus calling init_page_info here feels
wrong.  New function in utils.c that performs all the needed changes,
and have just one make_cleanup_restore_foo routine to switch back?