<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    Hi Hussain,<br>
    <br>
    have you upgraded to the latest Stackless version?<br>
    I think the necessary changes have been incorporated, lately.<br>
    <br>
    This function should not be visible from Python. It is just<br>
    a change that makes sys._getframe behave correctly.<br>
    <br>
    cheers - chris<br>
    <br>
    On 10/10/11 3:55 PM, Hussain Bohra wrote:
    <blockquote
      cite="mid:1318254900.9865.YahooMailNeo@web114713.mail.gq1.yahoo.com"
      type="cite">
      <div style="color:#000; background-color:#fff; font-family:Courier
        New, courier, monaco, monospace, sans-serif;font-size:10pt">
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2">Hi Christian,</font></span></div>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2"><br>
            </font></span></div>
        <div><font class="Apple-style-span" face="arial, helvetica,
            sans-serif" size="2"><span>I have incorporated the following
              changes in the </span>file<b> stackless_util.c</b>:</font></div>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2"><i><br>
              </i></font></span></div>
        <blockquote class="webkit-indent-blockquote"
          style="border-top-style: none; border-right-style: none;
          border-bottom-style: none; border-left-style: none;
          border-width: initial; border-color: initial; padding-top:
          0px; padding-right: 0px; padding-bottom: 0px; padding-left:
          0px; ">
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>static PyObject *</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>tasklet_get_frame(PyTaskletObject
                  *task)</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>{</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    PyFrameObject *f
                  = (PyFrameObject *) slp_get_frame(task);</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    PyObject *ret;</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i><br>
                </i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    while (f != NULL
                  && !PyFrame_Check(f)) {</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>            f =
                  f->f_back;</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    }</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    ret = (PyObject
                  *) f;</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i><br>
                </i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    if (ret == NULL)
                  ret = Py_None;</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    Py_INCREF(ret);</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>    return ret;</i></font></span></div>
          <div><span><font class="Apple-style-span" face="arial,
                helvetica, sans-serif" size="2"><i>}</i></font></span></div>
        </blockquote>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2"><br>
            </font></span></div>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2">I am able to build the
              stackless-python without any errors.</font></span></div>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2">But after which I am not
              able to use this method from the python prompt, I am not
              sure whether I need to write any specific code to expose
              this new method?</font></span></div>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2"><br>
            </font></span></div>
        <div><span><font class="Apple-style-span" face="arial,
              helvetica, sans-serif" size="2">Can you please let me know
              your feedback.</font></span></div>
        <div><font class="Apple-style-span" face="arial, helvetica,
            sans-serif" size="2"><br>
          </font></div>
        <div><font class="Apple-style-span" face="arial, helvetica,
            sans-serif" size="2">Thanks in advance.</font></div>
        <div><font class="Apple-style-span" face="arial, helvetica,
            sans-serif" size="2">Hussain Bohra</font></div>
        <div style="font-family: 'Courier New', courier, monaco,
          monospace, sans-serif; font-size: 10pt; "><br>
        </div>
        <div style="font-size: 10pt; font-family: 'Courier New',
          courier, monaco, monospace, sans-serif; ">
          <div style="font-size: 12pt; font-family: 'times new roman',
            'new york', times, serif; "><font face="Arial" size="2">
              <hr size="1"><b><span style="font-weight:bold;">From:</span></b>
              Hussain Bohra <a class="moz-txt-link-rfc2396E" href="mailto:hussainbohra_30@yahoo.com"><hussainbohra_30@yahoo.com></a><br>
              <b><span style="font-weight: bold;">To:</span></b>
              <a class="moz-txt-link-rfc2396E" href="mailto:stackless@stackless.com">"stackless@stackless.com"</a> <a class="moz-txt-link-rfc2396E" href="mailto:stackless@stackless.com"><stackless@stackless.com></a><br>
              <b><span style="font-weight: bold;">Sent:</span></b>
              Thursday, 11 August 2011 3:23 PM<br>
              <b><span style="font-weight: bold;">Subject:</span></b>
              Re: How do I get a callstack of a Tasklet ?<br>
            </font><br>
            <div id="yiv675152500">
              <div>
                <div style="color: rgb(0, 0, 0); background-color:
                  rgb(255, 255, 255); font-size: 10pt; font-family:
                  arial, helvetica, sans-serif; ">
                  <div><span>Hi Christian,</span></div>
                  <div><span><br>
                    </span></div>
                  <div><span>Thanks for this solution.</span></div>
                  <div><span>I will incorporate changes suggested by you
                      and will let you know the updates.</span></div>
                  <div> </div>
                  <div style="color: rgb(0, 0, 0); text-align: left;
                    font-family: 'monotype corsiva'; "
                    id="yiv675152500RTEContent">
                    <div style="color: rgb(0, 0, 255); font-family:
                      'times new roman'; "><font size="3">Thanks and
                        Regards,<br>
                        Hussain Bohra<br>
                      </font></div>
                  </div>
                  <div><br>
                  </div>
                  <div style="font-size: 10pt; font-family: arial,
                    helvetica, sans-serif; ">
                    <div style="font-size: 12pt; font-family: times,
                      serif; "><font face="Arial" size="2">
                        <hr size="1"><b><span style="font-weight:bold;">From:</span></b>
                        <a class="moz-txt-link-rfc2396E" href="mailto:stackless-request@stackless.com">"stackless-request@stackless.com"</a>
                        <a class="moz-txt-link-rfc2396E" href="mailto:stackless-request@stackless.com"><stackless-request@stackless.com></a><br>
                        <b><span style="font-weight:bold;">To:</span></b>
                        <a class="moz-txt-link-abbreviated" href="mailto:stackless@stackless.com">stackless@stackless.com</a><br>
                        <b><span style="font-weight:bold;">Sent:</span></b>
                        Wednesday, 10 August 2011 3:30 PM<br>
                        <b><span style="font-weight:bold;">Subject:</span></b>
                        Stackless Digest, Vol 91, Issue 5<br>
                      </font><br>
                      Send Stackless mailing list submissions to<br>
                          <a moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:stackless@stackless.com"
                        target="_blank"
                        href="mailto:stackless@stackless.com">stackless@stackless.com</a><br>
                      <br>
                      To subscribe or unsubscribe via the World Wide
                      Web, visit<br>
                          <a moz-do-not-send="true" rel="nofollow"
                        target="_blank"
                        href="http://www.stackless.com/mailman/listinfo/stackless">http://www.stackless.com/mailman/listinfo/stackless</a><br>
                      or, via email, send a message with subject or body
                      'help' to<br>
                          <a moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:stackless-request@stackless.com"
                        target="_blank"
                        href="mailto:stackless-request@stackless.com">stackless-request@stackless.com</a><br>
                      <br>
                      You can reach the person managing the list at<br>
                          <a moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:stackless-owner@stackless.com"
                        target="_blank"
                        href="mailto:stackless-owner@stackless.com">stackless-owner@stackless.com</a><br>
                      <br>
                      When replying, please edit your Subject line so it
                      is more specific<br>
                      than "Re: Contents of Stackless digest..."<br>
                      <br>
                      <br>
                      Today's Topics:<br>
                      <br>
                        1. pickling bug    (was:  Stackless Digest, Vol
                      90, Issue 3)<br>
                            (Christian Tismer)<br>
                        2. About submissions to this list (Christian
                      Tismer)<br>
                        3. Re: About submissions to this list (Rob
                      Coops)<br>
                        4. Re: About submissions to this list (Richard
                      Tew)<br>
                        5. Re: About submissions to this list (Christian
                      Tismer)<br>
                      <br>
                      <br>
----------------------------------------------------------------------<br>
                      <br>
                      Message: 1<br>
                      Date: Tue, 09 Aug 2011 16:06:05 +0200<br>
                      From: Christian Tismer <<a
                        moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:tismer@stackless.com"
                        target="_blank"
                        href="mailto:tismer@stackless.com">tismer@stackless.com</a>><br>
                      To: The Stackless Python Mailing List <<a
                        moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:stackless@stackless.com"
                        target="_blank"
                        href="mailto:stackless@stackless.com">stackless@stackless.com</a>><br>
                      Cc: Kristj?n Valur J?nsson     <<a
                        moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:kristjan@ccpgames.com"
                        target="_blank"
                        href="mailto:kristjan@ccpgames.com">kristjan@ccpgames.com</a>>,
                      Richard Tew<br>
                          <<a moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:richard.m.tew@gmail.com"
                        target="_blank"
                        href="mailto:richard.m.tew@gmail.com">richard.m.tew@gmail.com</a>><br>
                      Subject: [Stackless] pickling bug    (was: 
                      Stackless Digest, Vol 90,<br>
                          Issue 3)<br>
                      Message-ID: <<a moz-do-not-send="true"
                        rel="nofollow"
                        ymailto="mailto:4E413ECD.5050709@stackless.com"
                        target="_blank"
                        href="mailto:4E413ECD.5050709@stackless.com">4E413ECD.5050709@stackless.com</a>><br>
                      Content-Type: text/plain; charset=ISO-8859-1;
                      format=flowed<br>
                      <br>
                      Hi,<br>
                      <br>
                      I sent this almost-patch a while ago, but it was
                      probably invisible,<br>
                      due to a bad subject line. Here it is again:<br>
                      <br>
                      file stackless_util.c:<br>
                      <br>
                      static PyObject *<br>
                      tasklet_get_frame(PyTaskletObject *task)<br>
                      {<br>
                          PyFrameObject *f = (PyFrameObject *)
                      slp_get_frame(task);<br>
                          PyObject *ret;<br>
                      <br>
                          while (f != NULL && !PyFrame_Check(f))
                      {<br>
                                  f = f->f_back;<br>
                          }<br>
                          ret = (PyObject *) f;<br>
                      <br>
                          if (ret == NULL) ret = Py_None;<br>
                          Py_INCREF(ret);<br>
                          return ret;<br>
                      }<br>
                      <br>
                      This should resolve the problem.<br>
                      <br>
                      ciao - chris<br>
                      <br>
                      On 7/8/11 2:26 PM, Christian Tismer wrote:<br>
                      > On 7/8/11 1:34 PM, Christian Tismer wrote:<br>
                      >> On 7/8/11 1:53 AM, Richard Tew wrote:<br>
                      >>> On Thu, Jul 7, 2011 at 10:10 PM,
                      Hussain Bohra<br>
                      >>> <<a moz-do-not-send="true"
                        rel="nofollow"
                        ymailto="mailto:hussainbohra_30@yahoo.com"
                        target="_blank"
                        href="mailto:hussainbohra_30@yahoo.com">hussainbohra_30@yahoo.com</a>> 
                      wrote:<br>
                      >>>> Hi Christian,<br>
                      >>>><br>
                      >>>> Thanks for your feedback.<br>
                      >>>><br>
                      >>>> Could you kindly let us know when
                      you are done with the changes, so <br>
                      >>>> that we<br>
                      >>>> can complete the existing tasks
                      that are dependent on this fix.<br>
                      >>> Can't you work around it?  Like check
                      if a frame is a cframe and skip<br>
                      >>> it, if it is the case?  I haven't
                      looked at the traceback generation<br>
                      >>> in detail, but I imagine that
                      something along the following lines<br>
                      >>> would work..<br>
                      >>><br>
                      >>> lastTasklet = stackless.current<br>
                      >>> thisTasklet = lastTasklet.next<br>
                      >>> while True:<br>
                      >>>      thisFrame = thisTasklet.frame<br>
                      >>>      while thisFrame is not None:<br>
                      >>>          if not isinstance(thisFrame,
                      stackless.cframe):<br>
                      >>>              # Do traceback related
                      stuff.<br>
                      >>>          thisFrame = thisFrame.f_back<br>
                      >>>      thisTasklet = thisTasklet.next<br>
                      >>>      if lastTasklet is thisTasklet:<br>
                      >>>          break<br>
                      >><br>
                      >> Hi Richard, Hussain,<br>
                      >><br>
                      >> yes, I had a look now, and you are right.
                      CFrames get into the way,<br>
                      >> but are a small problem. Your workaround
                      should be a simple quick<br>
                      >> solution in the first place, and I
                      encourage Hussain to try that first.<br>
                      >> It is a really simple patch to
                      traceback.py: insert a check into <br>
                      >> extract_stack<br>
                      >> before "if limit is None:":<br>
                      >><br>
                      >>    while f and not isinstance(f,
                      types.FrameType):<br>
                      >>        f = f.f_back<br>
                      >><br>
                      >> That should do.<br>
                      >><br>
                      >>
                      ---------------------------------------------<br>
                      >> Now to the omission in stackless:<br>
                      >> The special treatment of cframes is quite
                      limited in stackless:<br>
                      >><br>
                      >> - sys._getframe has a patch to skip over
                      anything that is no normal <br>
                      >> frame.<br>
                      >> - frame objects have a getter function
                      frame_getback() instead of the <br>
                      >> f_back<br>
                      >>    field.<br>
                      >><br>
                      >> The only way how the treatment of cframes
                      might be circumvented<br>
                      >> in extract_stack is when a cframe is
                      given as an argument, and that<br>
                      >> was the omission in stackless:<br>
                      >> tasklet.frame exposes whatever it has to
                      the user. Only for the running<br>
                      >> tasklet, this gets redirected to the
                      threadstate's frame, which in the<br>
                      >> from python visible context never is a
                      cframe.<br>
                      >><br>
                      >> The function slp_get_frame is the culprit
                      (stackless_util.c):<br>
                      >><br>
                      >> /* CAUTION: This function returns a
                      borrowed reference */<br>
                      >> PyFrameObject *<br>
                      >> slp_get_frame(PyTaskletObject *task)<br>
                      >> {<br>
                      >>    PyThreadState *ts =
                      PyThreadState_GET();<br>
                      >><br>
                      >>    return ts->st.current == task ?
                      ts->frame : task->f.frame;<br>
                      >> }<br>
                      >><br>
                      >> This is correct for builtin API
                      functions, but not for the python<br>
                      >> interface. The getter methon
                      tasklet_get_frame (taskletobject.c)<br>
                      >> should skip over cframes.<br>
                      >><br>
                      >> static PyObject *<br>
                      >> tasklet_get_frame(PyTaskletObject *task)<br>
                      >> {<br>
                      >>    PyObject *ret = (PyObject*)
                      slp_get_frame(task);<br>
                      >><br>
                      >>    if (ret == NULL) ret = Py_None;<br>
                      >>    Py_INCREF(ret);<br>
                      >>    return ret;<br>
                      >> }<br>
                      >><br>
                      >> I think something like the following
                      should do it:<br>
                      >><br>
                      >> static PyObject *<br>
                      >> tasklet_get_frame(PyTaskletObject *task)<br>
                      >> {<br>
                      >>    PyFrameObject *f = (PyFrameObject *)
                      slp_get_frame(task);<br>
                      >>    PyObject *ret;<br>
                      >><br>
                      >>    while (f != NULL &&
                      !PyFrame_Check(f)) {<br>
                      >>            f = f_fback;<br>
                      >>    }<br>
                      >>    ret = (PyObject *) f;<br>
                      >><br>
                      >>    if (ret == NULL) ret = Py_None;<br>
                      >>    Py_INCREF(ret);<br>
                      >>    return ret;<br>
                      >> }<br>
                      >><br>
                      >><br>
                      >> Not yet tested, but I guess that should
                      do it.<br>
                      >> Do you want to try when removing
                      tasklet_become, or should I?<br>
                      >><br>
                      >> cheers -- chris<br>
                      >><br>
                      ><br>
                      > Correction:<br>
                      >            f = f_fback;<br>
                      > should be<br>
                      >            f = f->f_back;<br>
                      ><br>
                      > And as a remark:<br>
                      > The long-term solution to such problems is to
                      get rid of these old <br>
                      > tricks,<br>
                      > minimizing the patches to cpython as much as
                      possible.<br>
                      > We should remove all additions/changes to
                      frames and replace<br>
                      > cframes by helper structures that are
                      associated with a frame<br>
                      > invisibly and indirectly through a
                      dictionary.<br>
                      > Of course a bit tedious, because frames do
                      not support weak references<br>
                      > (<a moz-do-not-send="true" rel="nofollow"
                        target="_blank"
                        href="http://docs.python.org/library/weakref.html">http://docs.python.org/library/weakref.html</a>).<br>
                      > I'm thinking of a shadow stack of thunks as
                      an execution context<br>
                      > which references the frames, but is used
                      instead of the frame stack.<br>
                      ><br>
                      > Maybe it's time for another stackless sprint,
                      to start a major <br>
                      > clean-up? :-)<br>
                      ><br>
                      > cheers - chris<br>
                      ><br>
                      <br>
                      <br>
                      -- <br>
                      Christian Tismer            :^)<mailto:<a
                        moz-do-not-send="true" rel="nofollow"
                        ymailto="mailto:tismer@stackless.com"
                        target="_blank"
                        href="mailto:tismer@stackless.com">tismer@stackless.com</a>><br>
                      tismerysoft GmbH            :    Have a break!
                      Take a ride on Python's<br>
                      Johannes-Niemeyer-Weg 9A    :    *Starship* <a
                        moz-do-not-send="true" rel="nofollow"
                        target="_blank"
                        href="http://starship.python.net/">http://starship.python.net/</a><br>
                      14109 Berlin                :    PGP key ->  <a
                        moz-do-not-send="true" rel="nofollow"
                        target="_blank" href="http://wwwkeys.pgp.net/">http://wwwkeys.pgp.net/</a><br>
                      work +49 30 802 86 56  mobile +49 173 24 18 776 
                      fax +49 30 80 90 57 05<br>
                      PGP 0x57F3BF04      9064 F4E1 D754 C2FF 1619  305B
                      C09C 5A3B 57F3 BF04<br>
                            whom do you want to sponsor today?  <a
                        moz-do-not-send="true" rel="nofollow"
                        target="_blank" href="http://www.stackless.com/">http://www.stackless.com/</a><br>
                      <br>
                    </div>
                  </div>
                </div>
              </div>
            </div>
            <br>
            <br>
          </div>
        </div>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
Stackless mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Stackless@stackless.com">Stackless@stackless.com</a>
<a class="moz-txt-link-freetext" href="http://www.stackless.com/mailman/listinfo/stackless">http://www.stackless.com/mailman/listinfo/stackless</a></pre>
    </blockquote>
    <br>
    <br>
    <pre class="moz-signature" cols="72">-- 
Christian Tismer             :^)   <a class="moz-txt-link-rfc2396E" href="mailto:tismer@stackless.com"><mailto:tismer@stackless.com></a>
tismerysoft GmbH             :     Have a break! Take a ride on Python's
Karl-Liebknecht-Str. 121     :    *Starship* <a class="moz-txt-link-freetext" href="http://starship.python.net/">http://starship.python.net/</a>
14482 Potsdam                :     PGP key -> <a class="moz-txt-link-freetext" href="http://pgp.uni-mainz.de">http://pgp.uni-mainz.de</a>
work +49 173 24 18 776  mobile +49 173 24 18 776  fax n.a.
PGP 0x57F3BF04       9064 F4E1 D754 C2FF 1619  305B C09C 5A3B 57F3 BF04
      whom do you want to sponsor today?   <a class="moz-txt-link-freetext" href="http://www.stackless.com/">http://www.stackless.com/</a></pre>
  </body>
</html>