On 2017/05/07 20:50:15 -0700, Paul E. McKenney wrote: > On Sun, May 07, 2017 at 09:23:03AM +0900, Akira Yokosawa wrote: >> On 2017/05/03 07:59:48 -0700, Paul E. McKenney wrote: >>> On Wed, May 03, 2017 at 05:14:57PM +0900, Akira Yokosawa wrote: >>>> On 2017/05/02 19:47:19 -0700, Paul E. McKenney wrote: >>>>> On Wed, May 03, 2017 at 09:23:11AM +0900, Akira Yokosawa wrote: >>>>>> On 2017/05/03 09:09:37 +0900, Akira Yokosawa wrote: >>>>>>> Hi Paul, >>>>>>> >>>>>>> While reviewing your recent update in "Hunting Heisenbugs" section, >>>>>>> I found the analogy with particle physics didn't sit well with me. >>>>>>> >>>>>>> "Heisenbug" is defined as: >>>>>>> >>>>>>>> attempts to track down the heisenbug causes it to radically change >>>>>>>> its symptoms or even disappear completely. >>>>>>> >>>>>>> "Anti-heisenbug" is introduced as: >>>>>>> >>>>>>>> Fortunately, particle physics is up to the task: Why not create >>>>>>>> an anti-heisenbug to annihilate the heisenbug? >>>>>>> >>>>>>> At this point, "anti-heisenbug" seems to mean a patch that fixes >>>>>>> the heisenbug. >>>>>>> >>>>>>> The use of anti-something in particle physics does not go along >>>>>>> with this usage. >>>>>>> >>>>>>> For example, antiproton is an antiparticle of proton. >>>>>>> antiproton has mostly the same property as proton, >>>>>>> with the exception of electric charge and magnetic moment. >>>>>>> Also, an antiproton and an antielectron (positron) can compose >>>>>>> an antihydrogen atom. Well, I'm sure you know all about this. >>>>>>> >>>>>>> So, if you employ this line of reasoning, an anti-heisenbug should >>>>>>> also be a bug with something opposite in nature. >>>>>>> >>>>>>> If a heisenbug is a bug which vanishes when it is being tracked down, >>>>>>> an anti-heisenbug would be a bug which only emerges when it (or >>>>>>> something else) is being tracked down. >>>>>>> >>>>>>> However, it is not obvious you can create an anti-heisenbug for >>>>>>> a particular heisenbug so that when the two are combined, the symptoms >>>>>>> of both bugs to "annihilate". >>>>>>> >>>>>>> So I'm wondering "What was Paul's intention in using the word 'annihilate'?". >>>>>>> >>>>>>> In Section "Add Delay", anti-heisenbug is used as: >>>>>>> >>>>>>>> Once you spot a bug involving a race condition, it is frequently >>>>>>>> possible to create an anti-heisenbug by adding delay in this manner. >>>>>>> >>>>>>> This "anti-heisenbug" is not a bug, rather, it is a means to increase >>>>>>> the possibility of a given heisenbug to appear. >>>>>>> >>>>>> >>>>>> Oops, I sent this prematurely. >>>>>> But you might be able to see my point here. >>>>>> >>>>>> This is not an important question, so please give it a look when you have >>>>>> done with whatever issue you are working on. >>>>> >>>>> I freely admit that I am taking a fair amount of "poetic license" here. ;-) >>>> >>>> So, what about adding a footnote like this? >>> >>> That isn't bad at all! >>> >>> But let's take a look at several possible approaches: >>> >>> o Continue using "anti-heisenbug", but remove the current definition. >>> (To the extent that it could be considered to be a definition.) >>> >>> o Use the following analogy: >>> >>> When you annihilate an electron and a positron, you don't >>> get nothingness, you instead get an energetic photon. >>> Similarly, when you annihilate a heisenbug and an >>> anti-heisenbug, you don't (usually) get nothingness, >>> instead, you get a (hopefully) easier-to-locate bug. >>> >>> o The solution you propose below. >>> >>> o Leave as is, and assume that people will get the "poetic license". >>> >>> o Additional ideas here! >>> >>> o Any of the above, but have a quick quiz on the definition >>> instead of a footnote or an inline definition/explanation. >>> >>> Thoughts? >> >> I looked around the use of "heisenbug" and found an entry in quora: >> >> https://www.quora.com/Why-do-people-mix-up-the-observer-effect-and-the-uncertainty-principle >> >> What "heisenbug" should imply is the "observer effect", not the "uncertainty principle". >> >> So the definition of heisenbug given in the text is already confusing. > > That may be, but I wasn't the one who defined heisenbug. ;-) > > But given that it is defined, and given that a very large number of people > are very scared of heisenbugs, it is my job to reduce, preferably remove, > the fear. Less fear is needed so that they can better deal with the > heisenbugs that they will inevitably come across. > >> An example of observer effect in quantum mechanics is interference of >> electron beam passing through double slit. >> If you don't know which slit each electron passes, you can observe interference >> pattern. But once you put some equipment to know which slit an individual >> electron passes, you no longer see the interference pattern. >> >> This situation is much relevant to heisenbug, isn't it? > > Well, it actually -might- be possible to come up with a heisenbug that > mimicked quantum superpositioning and entanglement, but I don't know of > one right offhand. Actually, it -has- to be possible to generate such a > heisenbug because we can simulate quantum systems with software. Whether > or not such a heisenbug would actually occur naturally, I couldn't say. > Seems a bit unlikely, but then again, so did RCU. ;-) Hmm, hard to imagine... > >> Heisenberg used observer effect to explain his uncertainty principle as >> a thought experiment on the gamma-ray microscope. >> This is the origin of the confusion. And most people are still confused. >> >> So I don't like to see further analogies with misleading implications, which >> would just look odd or strange for those familiar with particle physics who >> happens to be also interested in parallel processing. > > The readers need to laugh enough, be confused enough, or whatever, in order > to let go of their fear. ;-) > >> All we need is to *somehow* make (the symptom of) a given heisenbug be >> reliably reproduced. And those "somehow"s are the theme of this section. > > That is true! > >> So, I'd suggest 1) fixing the definition of heisenbug, 2) departing from >> the analogy of particle physics. One problem is what name to use instead of >> anti-heisenbug. If you use "anti-" as in "antibiotics", anti-heisenbug >> is not necessarily confusing. > > I am not at all optimistic about being able to fix the definition > of heisenbug. It has been in use for far too long. And most people are likely to have heard of the uncertainty principle. > >> Or you might be able to come up with a better analogy by looking into studies >> in quantum measurement such as "weak measurement". This field is related to >> quantum computing you are planning to include into perfbook. >> >> See e.g. http://iopscience.iop.org/article/10.1088/0031-8949/91/7/073001/meta >> >> This is a review paper written from experimenters' POV, and should be easier >> to read than theoretical papers cited therein, I suppose. > > But if we have to educate the typical reader about weak measurement, we > have already lost. Point taken. > We will lose them, and they will retain their > crippling fear of heisenbugs. > > How about something like the following patch? Fair enough. And "heisen-ness" sounds funny enough. If you want to add, Acked-by: Akira Yokosawa <akiyks@xxxxxxxxx> Thanks, Akira > > Thanx, Paul > > ---------------------------------------------------------------------- > > commit 50d1f21f54540ff6ec38cc44715c213c3e16f2fe > Author: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx> > Date: Sun May 7 20:46:19 2017 -0700 > > debugging: Add footnote stating that "heisenbug" is a misnomer > > https://www.quora.com/Why-do-people-mix-up-the-observer-effect-and-the-uncertainty-principle > http://iopscience.iop.org/article/10.1088/0031-8949/91/7/073001/meta > > But the word is well entrenched in common usage in parallel programming. > > Reported-by: Akira Yokosawa <akiyks@xxxxxxxxx> > Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx> > > diff --git a/debugging/debugging.tex b/debugging/debugging.tex > index e621ebdb540b..c41057f35c31 100644 > --- a/debugging/debugging.tex > +++ b/debugging/debugging.tex > @@ -1370,23 +1370,30 @@ The next section discusses counter-intuitive ways of improving this situation. > > This line of thought also helps explain heisenbugs: > adding tracing and assertions can easily reduce the probability > -of a bug appearing. > -And this is why extremely lightweight tracing and assertion mechanism are > +of a bug appearing, which > +is why extremely lightweight tracing and assertion mechanism are > so critically important. > > -The name ``heisenbug'' stems from the Heisenberg Uncertainty Principle > -from quantum physics, which states that it is impossible to exactly > -quantify a given particle's position and velocity at any given point > -in time~\cite{WeinerHeisenberg1927Uncertain}. > +The term ``heisenbug'' was inspired by the Heisenberg Uncertainty > +Principle from quantum physics, which states that it is impossible to > +exactly quantify a given particle's position and velocity at any given > +point in time~\cite{WeinerHeisenberg1927Uncertain}. > Any attempt to more accurately measure that particle's position will > result in increased uncertainty of its velocity. > -A similar effect occurs for heisenbugs: attempts to track down the heisenbug > -causes it to radically change its symptoms or even disappear completely. > +A roughly similar effect occurs for heisenbugs: attempts to track down > +the heisenbug causes it to radically change its symptoms or even disappear > +completely.\footnote{ > + The term ``heisenbug'' is a misnomer, as most heisenbugs are > + fully explained by the \emph{observer effect} from classical > + physics. > + Nevertheless, the name ``heisenbug'' has stuck.} > > If the field of physics inspired the name of this problem, it is only > -logical that we should look to the field of physics for the solution. > +logical that the field of physics should inspire the solution. > Fortunately, particle physics is up to the task: > Why not create an anti-heisenbug to annihilate the heisenbug? > +Or, perhaps more accurately, to annihilate the heisen-ness of > +the heisenbug? > > This section describes a number of ways to do just that: > > > -- To unsubscribe from this list: send the line "unsubscribe perfbook" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
