Debug tracing

[mdblack98]

See if we can provide a "level" indication in the debug messages to show the function call depth. Would make reviewing the debug info easier to trace.

So level 1 at top and each level down would increment by 1 on entry and decrement by 1 on exit.
This can probably be done via the ENTERFUNC and RETURNFUNC. But RETURNFUNC may be called without an associated ENTERFUNC and vice versa so a state machine will be needed to determine when to increment/decrement.
ENTERFUNC would be if (state==RETURNFUNC) ++depth
RETURNFUNC would be if (state == ENTERRFUNC) ++ depth
ENTERFUNC would set ENTERFUNC state
RETURNFUNC would set RETURNFUNC state
get_depth function/macro would return depth when state==ENTERFUNC and depth+1 when state==RETURNFUNC
ENTERFUNC would only increment depth if state==RETURNFUNC
RETURNFUNC would only decrement depth if state==ENTERFUNC

[gvanem]

And all instances of rig_debug(RIG_DEBUG_VERBOSE, "%s: _init called\n", __func__);
should IMHO be replaced with ENTERFUNC2 (or better yet ENTERFUNC2()).

I find that there is a huge amount of code being run when debug is disabled! E.g. in rig_send_raw(), the preprocessor
on clang-cl spits out these lined (edited and Astyled):

  do // for ENTERFUNC
  {
    ++rig->state.depth;

    do
    {
      strcpy (debugmsgsave2, debugmsgsave);
      snprintf (debugmsgsave, sizeof (debugmsgsave), "%.*s%d:%s(%d):%s entered\n", rig->state.depth, spaces(), rig->state.depth, (strrchr ("src/rig.c",
                '/') ? strrchrfile, '/') + 1 : "src/rig.c"), 7285, __func__);
      (rig_debug) (RIG_DEBUG_VERBOSE, "%.*s%d:%s(%d):%s entered\n", rig->state.depth, spaces(), rig->state.depth, (strrchr ("src/rig.c", '/') ? strrchrfile,
                   '/') + 1 : "src/rig.c"), 7285, __func__);
    }
    while (0);
  }
  while (0);


  int retval = write_block_sync (&rs->rigport, send, send_len);

  if (retval < 0)
  {
    do  // For rig_debug():
    {
      strcpy (debugmsgsave2, debugmsgsave);
      snprintf (debugmsgsave, sizeof (debugmsgsave), "%s: write_block_sync() failed, result=%d\n", __func__, retval);
      (rig_debug) (RIG_DEBUG_ERR, "%s: write_block_sync() failed, result=%d\n", __func__, retval);
    }
    while (0);
  }

[mdblack98]

This provides intelligent debug information for apps like WSJTX/JTDX and others when errors occur. The strerror() function returns the debug history even if it's not turned on.
At this point I don't intend to eliminate this.

[gvanem]

... function returns the debug history even if it's not turned on.

Into debugmsgsave[]? But what are debugmsgsave2[] and debugmsgsave3[] used for?
But I think strerror() is wrongly used on Windows in many places.

[mdblack98]

save2 and save3 are deprecated.  Will be removed eventually. Where do think strerror is used incorrectly?  Quite possible... On Monday, June 6, 2022, 08:42:44 AM CDT, Gisle Vanem ***@***.***> wrote: ... function returns the debug history even if it's not turned on. Into debugmsgsave[]? But what are debugmsgsave2[] and debugmsgsave3[] used for? But I think strerror() is wrongly used on Windows in many places. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you modified the open/close state.Message ID: ***@***.***>

[gvanem]

Where do think strerror is used incorrectly?

2 places in src/network.c for example.

[mdblack98]

I found one and fixed it. Which two did you find? Mike On Monday, June 6, 2022, 09:17:44 AM CDT, Gisle Vanem ***@***.***> wrote: Where do think strerror is used incorrectly? 2 places in src/network.c for example. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you modified the open/close state.Message ID: ***@***.***>