AnsweredAssumed Answered

FW: vrf FW: Atomic_Operation

Question asked by VRFuser on Nov 9, 1999
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"><HTML><HEAD><META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1"><META NAME="Generator" CONTENT="MS Exchange Server version 5.5.2448.0"><TITLE>FW: vrf FW: Atomic_Operation</TITLE></HEAD><BODY><BR><BR><P><FONT SIZE=2>-----Original Message-----</FONT><BR><FONT SIZE=2>From: Greg Lawson </FONT><BR><FONT SIZE=2>Sent: Wednesday, November 10, 1999 9:41 AM</FONT><BR><FONT SIZE=2>To: 'Bombich, Anthony A WES'</FONT><BR><FONT SIZE=2>Subject: RE: vrf FW: Atomic_Operation</FONT></P><BR><P><FONT SIZE=2>I can't really comment on that.  This is the only correspondence I had with Greg Goebel and I was satisfied with the answer in relationship to my original inquiry.</FONT></P><P><FONT SIZE=2>I think you have remember that VEE has competition, so I don't think they can afford to stand still.  I think his point was that under their current development scheme changing this operation or providing a method to prioritize threads was not being considered.  Therefore, we could rely on the current operation to stay intact.</FONT></P><P><FONT SIZE=2>-----Original Message-----</FONT><BR><FONT SIZE=2>From: Bombich, Anthony A WES [<A HREF="mailto:BOMBICA@wes.army.mil">mailto:BOMBICA@wes.army.mil</A>]</FONT><BR><FONT SIZE=2>Sent: Wednesday, November 10, 1999 9:25 AM</FONT><BR><FONT SIZE=2>To: 'Greg Lawson'</FONT><BR><FONT SIZE=2>Subject: RE: vrf FW: Atomic_Operation</FONT></P><BR><P><FONT SIZE=2>Greg,</FONT><BR><FONT SIZE=2> </FONT><BR><FONT SIZE=2>I picked up something in Greg Goebel's response below that is somewhat</FONT><BR><FONT SIZE=2>disturbing.  Am I understanding correctly that there is a reduced corporate</FONT><BR><FONT SIZE=2>emphasis on significant improvements to VEE in the future?   I did not think</FONT><BR><FONT SIZE=2>that VEE was a FULLY mature package yet.   If so, this is moving in a</FONT><BR><FONT SIZE=2>similar direction as HP Basic was takena few years back.  This is not</FONT><BR><FONT SIZE=2>something I want to hear.  Please comment.</FONT><BR><FONT SIZE=2> </FONT><BR><FONT SIZE=2>Anthony A. Bombich, Research Physicist</FONT><BR><FONT SIZE=2>US Army Engineer Research and Development Center</FONT><BR><FONT SIZE=2>Waterways Experiment Station</FONT><BR><FONT SIZE=2>Vicksburg, MS 39180</FONT><BR><FONT SIZE=2> </FONT><BR><FONT SIZE=2> </FONT><BR><FONT SIZE=2> -----Original Message-----</FONT><BR><FONT SIZE=2>From: Greg Lawson [<A HREF="mailto:GLAWSON@coinco.com">mailto:GLAWSON@coinco.com</A>]</FONT><BR><FONT SIZE=2>Sent: Wednesday, November 10, 1999 8:26 AM</FONT><BR><FONT SIZE=2>To: vrf@lvld.hp.com</FONT><BR><FONT SIZE=2>Subject: vrf FW: Atomic_Operation</FONT></P><BR><BR><P><FONT SIZE=2>There seems to a lot of discussion lately about the formula control pin.  We</FONT><BR><FONT SIZE=2>rely on this "trick" because it is the only method to control multi-threaded</FONT><BR><FONT SIZE=2>operation.  This is important if you have an interrupt (SRQ) that needs to</FONT><BR><FONT SIZE=2>be serviced without being disturbed by other threads.  In fact, other</FONT><BR><FONT SIZE=2>threads are relying on the interrupt handler to keep data current.</FONT></P><P><FONT SIZE=2>Because this is the only method for controlling multi-threading and we rely</FONT><BR><FONT SIZE=2>on it heavily, we asked HP through our field rep to indicate their position</FONT><BR><FONT SIZE=2>on this.  Is it considered a bug?  Will this operation remain consistent in</FONT><BR><FONT SIZE=2>future revisions?  If they modified this operation we wanted some other</FONT><BR><FONT SIZE=2>means of holding off other threads.</FONT></P><P><FONT SIZE=2>It ended up that Greg Goebel was assigned the task of responding to our</FONT><BR><FONT SIZE=2>inquiry.  Below is his response, I hope this helps every one out there in</FONT><BR><FONT SIZE=2>VEE world.</FONT></P><P><FONT SIZE=2>-----Original Message----- </FONT><BR><FONT SIZE=2>From: Greg Goebel [ <A HREF="mailto:gvg@hpislsup.lvld.hp.com">mailto:gvg@hpislsup.lvld.hp.com</A></FONT><BR><FONT SIZE=2>|<A HREF="mailto:gvg@hpislsup.lvld.hp.com">mailto:gvg@hpislsup.lvld.hp.com</A>> ] </FONT><BR><FONT SIZE=2>Sent: Wednesday, March 31, 1999 4:35 PM </FONT><BR><FONT SIZE=2>To: GLAWSON@Coinco.com </FONT><BR><FONT SIZE=2>Subject: Atomic_Operation </FONT></P><BR><P><FONT SIZE=2>send q Atomic_Operation ww nguyen_m GLAWSON@Coinco.com </FONT></P><P><FONT SIZE=2>from: Greg Goebel / HP-MXD </FONT><BR><FONT SIZE=2>      greg_goebel@hp.com / 800-452-4844 </FONT><BR><FONT SIZE=2>      website:  <A HREF="ftp://fcext3.external.hp.com/dist/mxd/index.html" TARGET="_blank">ftp://fcext3.external.hp.com/dist/mxd/index.html</A></FONT><BR><FONT SIZE=2>|<A HREF="ftp://fcext3.external.hp.com/dist/mxd/index.html" TARGET="_blank">ftp://fcext3.external.hp.com/dist/mxd/index.html</A>>  </FONT><BR><FONT SIZE=2>to:   Greg Lawson </FONT><BR><FONT SIZE=2>cc:   Minh Nguyen </FONT><BR><FONT SIZE=2>date: Wednesday, 31 March 1999 1521 MST </FONT></P><P><FONT SIZE=2>Sir: </FONT></P><P><FONT SIZE=2>I think the question boils down to the need to guarantee "atomic" </FONT><BR><FONT SIZE=2>(indivisible) operation of a specific VEE UserFunction. </FONT></P><P><FONT SIZE=2>The trick using the control pin does indeed provide atomic operation.  It </FONT><BR><FONT SIZE=2>is not documented.  I asked the lab person if this class of operation was </FONT><BR><FONT SIZE=2>guaranteed in the future. </FONT></P><P><FONT SIZE=2>The answer is a qualified yes.  In fact, the behavior you are relying on is </FONT><BR><FONT SIZE=2>ingrained into the regression tests for VEE and changing it would destroy </FONT><BR><FONT SIZE=2>our regression tests.  Furthermore, some other customers have complained</FONT><BR><FONT SIZE=2>that </FONT><BR><FONT SIZE=2>the behavior you require is a bug and should be fixed, and the lab had to </FONT><BR><FONT SIZE=2>respond that doing so would be much too difficult. </FONT></P><P><FONT SIZE=2>So formally speaking the status of this behavior is ambiguous, informally </FONT><BR><FONT SIZE=2>speaking it is unlikely to be changed.  This is all more the case because </FONT><BR><FONT SIZE=2>in the current strategy, only relatively modest changes are likely to be </FONT><BR><FONT SIZE=2>made to VEE itself.  We will extend VEE by adding ActiveX controls that can </FONT><BR><FONT SIZE=2>be leveraged to other products and make better use of our lab resources. </FONT></P><P><FONT SIZE=2>I cannot guarantee that this feature will never be changed, but in practice </FONT><BR><FONT SIZE=2>there is little that I can guarantee in any case.  However, I would take </FONT><BR><FONT SIZE=2>ten to one odds that this behavior will remain constant. </FONT></P><P><FONT SIZE=2>The atomic operation issue has come up enough so that the lab people once </FONT><BR><FONT SIZE=2>thought of actually adding a flag to UserFunctions to guarantee it.</FONT><BR><FONT SIZE=2>However, </FONT><BR><FONT SIZE=2>the change in VEE development focus makes it almost more certain this</FONT><BR><FONT SIZE=2>measure </FONT><BR><FONT SIZE=2>will never be done. </FONT></P><P><FONT SIZE=2>Am I on track with this response?  </FONT></P><P><FONT SIZE=2>[|>] regards -- gvg </FONT><BR><FONT SIZE=2>  </FONT></P></BODY></HTML>  

Outcomes