Loading ...
Sorry, an error occurred while loading the content.

Re: [XP] meaning of "could possibly work" clause

Expand Messages
  • Phlip
    ... Pass a test by any means necessary. Clone an entire class and change one line, if that s the most expedient way. If you can t find a simple way to pass the
    Message 1 of 22 , Jun 1, 2006
    • 0 Attachment
      Paul Campbell wrote:

      > Heres a question for the house: what does "simplest thing that could
      > possibly work" mean over and above just "simplest thing" ?. All it
      > seems to do is exclude things which dont work, and what is the point
      > of that in a TDD world ?

      Pass a test by any means necessary. Clone an entire class and change
      one line, if that's the most expedient way.

      If you can't find a simple way to pass the test, step back from the
      situation and look at it agan. Maybe you can sandbox and learn things,
      or maybe you can write a narrower and simpler test.

      After the test passes, you are free to refactor. Now simple means
      "elegant". Perform Extract Class Refactor on your two classes, to
      produce a new base class or delegate class.

      You can also pass a test by lying - by returning a hard-coded value,
      or some such. You fix that by writing enough test cases to force out
      the lie. And that provides cross-coverage.

      --
      Phlip
      http://c2.com/cgi/wiki?ZeekLand <-- NOT a blog!!
    • Ron Jeffries
      ... These are neat techniques and I use and recommend them all. In the present lack of context, I d be concerned that they all seem crazy. Ron Jeffries
      Message 2 of 22 , Jun 1, 2006
      • 0 Attachment
        On Thursday, June 1, 2006, at 7:52:26 AM, Phlip wrote:

        > Paul Campbell wrote:

        >> Heres a question for the house: what does "simplest thing that could
        >> possibly work" mean over and above just "simplest thing" ?. All it
        >> seems to do is exclude things which dont work, and what is the point
        >> of that in a TDD world ?

        > Pass a test by any means necessary. Clone an entire class and change
        > one line, if that's the most expedient way.

        > If you can't find a simple way to pass the test, step back from the
        > situation and look at it agan. Maybe you can sandbox and learn things,
        > or maybe you can write a narrower and simpler test.

        > After the test passes, you are free to refactor. Now simple means
        > "elegant". Perform Extract Class Refactor on your two classes, to
        > produce a new base class or delegate class.

        > You can also pass a test by lying - by returning a hard-coded value,
        > or some such. You fix that by writing enough test cases to force out
        > the lie. And that provides cross-coverage.

        These are neat techniques and I use and recommend them all. In the
        present lack of context, I'd be concerned that they all seem crazy.

        Ron Jeffries
        www.XProgramming.com
        This is how I develop software.
        Take the parts that make sense to you.
        Ignore the rest.
      • Paul Campbell
        ... But that base is covered by simplest on its own - the clause that could possibly work doesnt provide any further hint that expediency is not what is
        Message 3 of 22 , Jun 1, 2006
        • 0 Attachment
          --- In extremeprogramming@yahoogroups.com, Phlip <phlip2005@...> wrote:
          >
          > Paul Campbell wrote:
          >
          > > Heres a question for the house: what does "simplest thing that could
          > > possibly work" mean over and above just "simplest thing" ?. All it
          > > seems to do is exclude things which dont work, and what is the point
          > > of that in a TDD world ?
          >
          > Pass a test by any means necessary. Clone an entire class and change
          > one line, if that's the most expedient way.

          But that base is covered by "simplest" on its own - the clause "that
          could possibly work" doesnt provide any further hint that expediency
          is not what is being sought.

          Paul.
        • Paul Campbell
          ... I still maintain that the works bit is a strong enough axiomatic constraint for all software dev that it doesnt need to be restated. ... To me it sounds
          Message 4 of 22 , Jun 1, 2006
          • 0 Attachment
            --- In extremeprogramming@yahoogroups.com, Ron Jeffries
            <ronjeffries@...> wrote:
            >
            > On Thursday, June 1, 2006, at 5:35:41 AM, Paul Campbell wrote:
            >
            > > Heres a question for the house: what does "simplest thing that could
            > > possibly work" mean over and above just "simplest thing" ?. All it
            > > seems to do is exclude things which dont work, and what is the point
            > > of that in a TDD world ?.
            >
            > If it couldn't possibly work, then one might not want to start out
            > in that direction.

            I still maintain that the "works" bit is a strong enough axiomatic
            constraint for all software dev that it doesnt need to be restated.

            >
            > Kent Beck used to ask "what is the simplest thing that could
            > possibly work," to help people get out of analysis paralysis and do
            > something.
            >
            > I began to say "/Do/ the simplest thing that could possibly work,"
            > for two reasons. First, the power of the idea is to do something
            > simple enough to get started with, and second, I find it to be a
            > strong move to start with something really simple, because often it
            > works just fine.
            >
            > > But worse than merely being tuatologous, to me it encourages
            > > missinterpretation of "simple" as easiest/quickest/most expedient and
            > > thus is not a phrase I would ever use myself to convey what I mean by
            > > "simplest".
            >
            > I can see how "simplest thing" could be misinterpreted in those
            > ways, but not why "could possibly work" would make it worse as you
            > seem to suggest.

            To me it sounds alot more absolute. Even missing out "possibly" makes
            it alot better i.e. "simplest thing that works". When I read the
            original phrase I get a vibe that the "possibly/possible" applies to
            "simple" as well as "works" and thus in some sense reads as "absolutey
            the simplest possible thing which works" which is too strong, too hard
            edged and absolute. Maybe its just me :-).

            Paul.
          • Phlip
            ... Do the simplest thing that could possibly work , at the macro scale, is about fitting features to technology. For example, I know a shop that implemented
            Message 5 of 22 , Jun 1, 2006
            • 0 Attachment
              Paul Campbell wrote:

              > > Pass a test by any means necessary. Clone an entire class and change
              > > one line, if that's the most expedient way.
              >
              > But that base is covered by "simplest" on its own - the clause "that
              > could possibly work" doesnt provide any further hint that expediency
              > is not what is being sought.

              "Do the simplest thing that could possibly work", at the macro scale,
              is about fitting features to technology. For example, I know a shop
              that implemented their own plumbing layer, and is now spending a lot
              of time researching CORBA. Time to transfer out one simplicity for
              another.

              "Do the simplest thing that could possibly work", at the micro scale
              means different things at different times. While the bar is red, the
              rules are different.

              --
              Phlip
              http://c2.com/cgi/wiki?ZeekLand <-- NOT a blog!!
            • Ron Jeffries
              ... Your privilege. The simplest thing, however, usually won t work. We wouldn t want to suggest that. So the question is, how simple. My answer is: the first
              Message 6 of 22 , Jun 1, 2006
              • 0 Attachment
                On Thursday, June 1, 2006, at 11:57:23 AM, Paul Campbell wrote:

                >> If it couldn't possibly work, then one might not want to start out
                >> in that direction.

                > I still maintain that the "works" bit is a strong enough axiomatic
                > constraint for all software dev that it doesnt need to be restated.

                Your privilege. The simplest thing, however, usually won't work. We
                wouldn't want to suggest that. So the question is, how simple. My
                answer is: the first simple thing that you aren't sure won't work.

                Ron Jeffries
                www.XProgramming.com
                The main reason that testing at the end of a development cycle finds
                problems is not that problems were put in near the end, it is that
                testing was put off until then.
              • Ron Jeffries
                ... Always a possibility, I suppose, however unlikely. It seems important to me to find the least non-stupid idea. The simplest thing is probably stupid,
                Message 7 of 22 , Jun 1, 2006
                • 0 Attachment
                  On Thursday, June 1, 2006, at 11:57:23 AM, Paul Campbell wrote:

                  >> I can see how "simplest thing" could be misinterpreted in those
                  >> ways, but not why "could possibly work" would make it worse as you
                  >> seem to suggest.

                  > To me it sounds alot more absolute. Even missing out "possibly" makes
                  > it alot better i.e. "simplest thing that works". When I read the
                  > original phrase I get a vibe that the "possibly/possible" applies to
                  > "simple" as well as "works" and thus in some sense reads as "absolutey
                  > the simplest possible thing which works" which is too strong, too hard
                  > edged and absolute. Maybe its just me :-).

                  Always a possibility, I suppose, however unlikely.

                  It seems important to me to find the "least" non-stupid idea. The
                  simplest thing is probably stupid, because it won't work. Other
                  simple things need to be selected among. I like to focus on the
                  simplest one that I can't kick out on the basis of its not working.

                  Ron Jeffries
                  www.XProgramming.com
                  There is no art without intention. -- Duke Ellington
                • Andrew McDonagh
                  ... Philps set of Simplest ways of making a test pass then refactoring the simple code into well designed code are great, I ve used the same kind of set for
                  Message 8 of 22 , Jun 1, 2006
                  • 0 Attachment
                    Paul Campbell wrote:
                    > Heres a question for the house: what does "simplest thing that could
                    > possibly work" mean over and above just "simplest thing" ?. All it
                    > seems to do is exclude things which dont work, and what is the point
                    > of that in a TDD world ?.
                    >
                    > But worse than merely being tuatologous, to me it encourages
                    > missinterpretation of "simple" as easiest/quickest/most expedient and
                    > thus is not a phrase I would ever use myself to convey what I mean by
                    > "simplest".
                    >
                    > Paul.
                    >
                    >
                    Philps set of 'Simplest ways of making a test pass then refactoring the
                    simple code into well designed code' are great, I've used the same kind
                    of set for a number of years now and i would encourage anyone new to TDD
                    to use them.

                    That said, I think its worth adding what for me 'Simplest thing...' does
                    NOT mean.

                    It does NOT mean being lazy or ignoring the techniques, designs,
                    pit-falls we have already learned.

                    Yet this is often the first thing I see people doing in the name of
                    DTSTTCPW.

                    One example i saw the other year was when a pair decided that creating a
                    Singleton was the 'Simplest thing...'. Sure it was a simple solution to
                    their problem. It allowed them to quickly speed through the story, as
                    they didn't need to modify the current design, they could simply invoke
                    the getInstance() method wherever and whenever they liked - great!

                    Errr no, not really, it created the start of a mess, because an up
                    coming iteration required that we'd need to create two instances of the
                    class.

                    Whilst working on this later iteration, one of the original pair was
                    also working on the story which required us to create a second instance
                    of the class. Again, the 'Simplest thing...' misconception kicked in
                    with them. They didn't look at the current singleton design and think
                    'Hmm we should refactor this singleton away'. No, they 'Simply' added a
                    setInstance() method to the Singleton.

                    So now we had a singleton that actually wasn't a singleton - all because
                    of they'd taken the 'Simplest thing...' too literally.

                    They even joked that they'd '..had to create a Doubleton' - its humor
                    was lost on the delay it caused when we had to change the design.

                    I'd agreed whole heartedly therefore that Simplest can often be mistaken
                    (whether deliberately or not) as easiest/quickest/most expedient.

                    Andrew
                  • Walter Prins
                    Hello Andew, Yes. To me the work part of could possibly works implies that you need to take a brief step back and ensure that whatever you come up with
                    Message 9 of 22 , Jun 1, 2006
                    • 0 Attachment
                      Hello Andew,

                      Yes. To me the "work" part of "could possibly works" implies that you
                      need to take a brief step back and ensure that whatever you come up with
                      must be able to at least satisfy not just the immediate need (that is,
                      to get to green for this story), but you must at least be able to also
                      foresee some simple-ish path forward for the other bits you're aware
                      that will happen or will be needed (as imperfect and incomplete as that
                      understanding might be at that moment.) At the very least, whatever
                      you do now in the name of simplicity must not obviously cause complexity
                      and trouble come the next story.

                      So, considering your Singleton example: for this reason, I would
                      probably have rejected the whole idea of using a singleton there as I
                      would've seen that it probably would not work (or, at least, not in a
                      simple manner) when we got to some of the other stories that were in the
                      pipeline.

                      It seems to me that interpreting TSTTCPW in this way thus helps avoid (I
                      hope/I think) the oversimplistic thinking that you refer to. Thoughts?

                      Regards,

                      Walter

                      Andrew McDonagh wrote:
                      >
                      > I'd agreed whole heartedly therefore that Simplest can often be mistaken
                      > (whether deliberately or not) as easiest/quickest/most expedient.
                      >
                      >
                    • Anthony Kaufman
                      I discovered XP about a year and a half ago and have been migrating my thinking in that direction since then so I m by no means an expert. I have to confess
                      Message 10 of 22 , Jun 8, 2006
                      • 0 Attachment
                        I discovered XP about a year and a half ago and have been migrating my
                        thinking in that direction since then so I'm by no means an expert. I have
                        to confess that I did initially take the simplest thing to mean the easiest
                        thing.

                        I'll go back to Kent's list in light of Andrew's issue.
                        1. Runs all the tests - The Singleton obviously ran all tests, so did the
                        Doubleton.
                        2. Contains no duplication - Guessing they both still pass this test as
                        well.
                        3. Expresses all the code's design ideas - The Singleton totally could have
                        done this but I don't know what the code actually looked like. However, the
                        Doubleton breaks down here. It sounds like it doesn't do a great job of
                        making the design clear.
                        4. Minimizes entities (classes, methods, ...) - Yeah, sounds like they both
                        achieve this.

                        The question in my mind now is:
                        Does the simplest design state that you should take into account stories to
                        come when constructing your design for the story right now?

                        If down the road you'll need a plane, but this story only wants something
                        that'll drive down the runway, should you construct landing gear and some
                        fuselage or just a build a moped?

                        "Right" kindof depends on how things turn out. If you end up constructing a
                        plane as was planned, you'll be pleased. Your extra effort to construct
                        landing gear and fuselage was worth it. However, if all you ever need is
                        something that'll drive down a runway, or if your planned "flying" stories
                        are scraped and replaced with "drive down the road" stories, you'll have
                        wasted time. And until you implement the flying stories you have this weird
                        plane shaped vehicle teetering around. Wasn't one of the goals of XP to
                        always have an ideal system at the end of an iteration?

                        Anthony

                        On 6/1/06, Walter Prins <wprins@...> wrote:
                        >
                        > Hello Andew,
                        >
                        > Yes. To me the "work" part of "could possibly works" implies that you
                        > need to take a brief step back and ensure that whatever you come up with
                        > must be able to at least satisfy not just the immediate need (that is,
                        > to get to green for this story), but you must at least be able to also
                        > foresee some simple-ish path forward for the other bits you're aware
                        > that will happen or will be needed (as imperfect and incomplete as that
                        > understanding might be at that moment.) At the very least, whatever
                        > you do now in the name of simplicity must not obviously cause complexity
                        > and trouble come the next story.
                        >
                        > So, considering your Singleton example: for this reason, I would
                        > probably have rejected the whole idea of using a singleton there as I
                        > would've seen that it probably would not work (or, at least, not in a
                        > simple manner) when we got to some of the other stories that were in the
                        > pipeline.
                        >
                        > It seems to me that interpreting TSTTCPW in this way thus helps avoid (I
                        > hope/I think) the oversimplistic thinking that you refer to. Thoughts?
                        >
                        > Regards,
                        >
                        > Walter
                        >
                        > Andrew McDonagh wrote:
                        > >
                        > > I'd agreed whole heartedly therefore that Simplest can often be mistaken
                        > > (whether deliberately or not) as easiest/quickest/most expedient.
                        > >
                        > >
                        >
                        >
                        > To Post a message, send it to: extremeprogramming@...
                        >
                        > To Unsubscribe, send a blank message to:
                        > extremeprogramming-unsubscribe@...
                        >
                        > ad-free courtesy of objectmentor.com
                        > Yahoo! Groups Links
                        >
                        >
                        >
                        >
                        >
                        >
                        >
                        >


                        [Non-text portions of this message have been removed]
                      • Ron Jeffries
                        ... No. It should run the tests, contain no duplication, express all the design ideas presently in it, and minize stuff ... ... A really good moped. ... Except
                        Message 11 of 22 , Jun 8, 2006
                        • 0 Attachment
                          On Thursday, June 8, 2006, at 3:31:05 PM, Anthony Kaufman wrote:

                          > Does the simplest design state that you should take into account stories to
                          > come when constructing your design for the story right now?

                          No. It should run the tests, contain no duplication, express all the
                          design ideas presently in it, and minize stuff ...

                          > If down the road you'll need a plane, but this story only wants something
                          > that'll drive down the runway, should you construct landing gear and some
                          > fuselage or just a build a moped?

                          A really good moped.

                          > "Right" kindof depends on how things turn out. If you end up constructing a
                          > plane as was planned, you'll be pleased. Your extra effort to construct
                          > landing gear and fuselage was worth it.

                          Except that it was waste investment doing it when all you were doing
                          was cruising around on the runway ... better to save the time and
                          money until later.

                          > However, if all you ever need is
                          > something that'll drive down a runway, or if your planned "flying" stories
                          > are scraped and replaced with "drive down the road" stories, you'll have
                          > wasted time. And until you implement the flying stories you have this weird
                          > plane shaped vehicle teetering around. Wasn't one of the goals of XP to
                          > always have an ideal system at the end of an iteration?

                          Ideal for what it then does. Not ideal for what it will someday do.
                          The rules of simple design make it ideal for what it has to do next.

                          Ron Jeffries
                          www.XProgramming.com
                          The work teaches us. -- Richard Gabriel
                        • geoffrey_slinker
                          ... wrote: ... Please help me out a bit. When I read these types of posts I wonder if the scenario in question is always contrived
                          Message 12 of 22 , Jun 8, 2006
                          • 0 Attachment
                            --- In extremeprogramming@yahoogroups.com, Ron Jeffries
                            <ronjeffries@...> wrote:
                            <snip>
                            > Ideal for what it then does. Not ideal for what it will someday do.
                            > The rules of simple design make it ideal for what it has to do next.
                            >

                            Please help me out a bit.

                            When I read these types of posts I wonder if the scenario in question
                            is always contrived with this idea that "we MAY need X, Y, or Z". It
                            almost sounds like all projects are small and no one incrementally
                            builds large systems, and no one knows the real customer needs.

                            Suppose the requirement says, "We need a vehicle that can go down a
                            runway". Firstly, the term runway is typically in the domain of
                            aeronautics and so one could ask, "Do you mean you need an airplane,
                            or do you need a firetruck, or do you need a baggage tractor?" But
                            maybe I am more familiar with the fashion industry and so I give you
                            a supermodel on roller blades.

                            These scenarios are often presented in such a way so that there are
                            no communication paths available to the developer.

                            For example, suppose the "product" to be built is an airplane. The
                            scenario is as if the Project Manager divied up the tasks to
                            different development teams that can not talk to each other and that
                            the big picture is only given on a need to know basis. It is like the
                            PM told one group, "Build me something that can roll down a runway"
                            and he told another group to "build me something that can seat 8
                            passengers" and yet a third group he told "build me something that
                            can make things fly".

                            There should be a big picture. You should know what is required. If
                            something comes up that you think may be needed then tell someone so
                            that it can be determined if something was overlooked. If it is not
                            an oversight then don't make it. Clearly simple design, regression
                            tests, properly defined domain, etc., can make the system easy to
                            extend and from my experience software is usually extended.

                            Many times I have heard, "xp doesn't mean check your brain at the
                            door". Making a big picture doesn't mean do BDUF. Understanding your
                            customer's requirements doesn't mean do BDUF. There is a planning
                            game to determing the product. Then this product (which is the big
                            picture) is divided up and releases and iterations are defined and
                            the product is delivered incrementally.

                            If you need a tractor, a plane, or a supermodel on roller blades, it
                            should be known by someone somewhere somehow.

                            I don't know, but it seems like I am missing something, or
                            misunderstanding something...

                            Geoff
                          • Andrew McDonagh
                            ... You are right, there is a big picture, there has been a release planning session and iteration sessions too. That unfortunately is not the issue. Its
                            Message 13 of 22 , Jun 8, 2006
                            • 0 Attachment
                              geoffrey_slinker wrote:
                              >
                              > --- In extremeprogramm
                              >
                              > If you need a tractor, a plane, or a supermodel on roller blades, it
                              > should be known by someone somewhere somehow.
                              >
                              > I don't know, but it seems like I am missing something, or
                              > misunderstanding something...
                              >
                              > Geoff


                              You are right, there is a big picture, there has been a release planning
                              session and iteration sessions too.

                              That unfortunately is not the issue. Its blind adherence to Do The
                              Simplest Thing for the CURRENT iteration.

                              When ever we create software, there are numerous ways we could do it -
                              some are more open to change than others. But this flexibility is often
                              not as Simple as the ridged design. And its the choosing to do the more
                              ridged design because its Simple that causes the problems.
                            • yahoogroups@jhrothjr.com
                              From: Andrew McDonagh To: extremeprogramming@yahoogroups.com
                              Message 14 of 22 , Jun 8, 2006
                              • 0 Attachment
                                From: "Andrew McDonagh"
                                <yahoogroups.at.andrewmcdonagh.f2s.com@...>
                                To: "extremeprogramming@yahoogroups.com"
                                <extremeprogramming.at.yahoogroups.com@...>
                                Sent: Thursday, June 08, 2006 5:32 PM
                                Subject: Re: [XP] meaning of "could possibly work" clause


                                > geoffrey_slinker wrote:
                                > >
                                > > --- In extremeprogramm
                                > >
                                > > If you need a tractor, a plane, or a supermodel on roller blades, it
                                > > should be known by someone somewhere somehow.
                                > >
                                > > I don't know, but it seems like I am missing something, or
                                > > misunderstanding something...
                                > >
                                > > Geoff
                                >
                                >
                                > You are right, there is a big picture, there has been a release planning
                                > session and iteration sessions too.
                                >
                                > That unfortunately is not the issue. Its blind adherence to Do The
                                > Simplest Thing for the CURRENT iteration.
                                >
                                > When ever we create software, there are numerous ways we could do it -
                                > some are more open to change than others. But this flexibility is often
                                > not as Simple as the ridged design. And its the choosing to do the more
                                > ridged design because its Simple that causes the problems.

                                Simplicity satisfies the _real_ requirements of the domain without
                                a lot of additional cruft, and without throwing all those embarrasing
                                complexities over the fence for someone else to deal with.

                                If you tell me that this code, once written, is never, ever going to
                                be maintained, I'm going to look at you with extreme skepticism.
                                In fact, if you can't give me a good reason why I should think that,
                                I'll look at you as either very naive or as lying to me in order to
                                manipulate me into not doing those things that I know make code
                                maintainable.

                                It doesn't matter to me how simple it looks, if it can't be maintained
                                it's not simple.

                                John Roth

                                What makes a design rigid?


                                >
                                >
                                >
                              • Ron Jeffries
                                ... I m not sure how to do a rigid design. I am sure that if the cod runs all the tests, has no duplication, expresses all the design ideas so far, and
                                Message 15 of 22 , Jun 8, 2006
                                • 0 Attachment
                                  On Thursday, June 8, 2006, at 7:32:18 PM, Andrew McDonagh wrote:

                                  > You are right, there is a big picture, there has been a release planning
                                  > session and iteration sessions too.

                                  > That unfortunately is not the issue. Its blind adherence to Do The
                                  > Simplest Thing for the CURRENT iteration.

                                  > When ever we create software, there are numerous ways we could do it -
                                  > some are more open to change than others. But this flexibility is often
                                  > not as Simple as the ridged design. And its the choosing to do the more
                                  > ridged design because its Simple that causes the problems.

                                  I'm not sure how to do a rigid design. I am sure that if the cod
                                  runs all the tests, has no duplication, expresses all the design
                                  ideas so far, and minimizes entities, it's bloody hard to also make
                                  it difficult to change.

                                  Yes, that is a challenge: I'd like to see some code that meets the
                                  Beck rules and is hard to change.

                                  Ron Jeffries
                                  www.XProgramming.com
                                  We know less about the project today than at any time in the future.
                                  -- Chet Hendrickson
                                  You mean today is the dumbest day of the rest of my life?
                                  -- Ron Jeffries
                                • Ron Jeffries
                                  ... I shall do my best. (Ben Stein) ... I m not assuming that we don t know, and I m not assuming that things may change. I assert that at any given moment, if
                                  Message 16 of 22 , Jun 8, 2006
                                  • 0 Attachment
                                    On Thursday, June 8, 2006, at 6:41:51 PM, geoffrey_slinker wrote:

                                    >> Ideal for what it then does. Not ideal for what it will someday do.
                                    >> The rules of simple design make it ideal for what it has to do next.
                                    >>

                                    > Please help me out a bit.

                                    I shall do my best. (Ben Stein)

                                    > When I read these types of posts I wonder if the scenario in question
                                    > is always contrived with this idea that "we MAY need X, Y, or Z". It
                                    > almost sounds like all projects are small and no one incrementally
                                    > builds large systems, and no one knows the real customer needs.

                                    I'm not assuming that we don't know, and I'm not assuming that
                                    things may change. I assert that at any given moment, if we're
                                    implementing feature by feature, the best way for the design to be
                                    is exactly right for the features then implemented, not exactly
                                    right for some future state.

                                    We might wish to explore what would have to be true for that
                                    assertion to be true, depending on where your thoughts, and those of
                                    others, go next.

                                    > Suppose the requirement says, "We need a vehicle that can go down a
                                    > runway". Firstly, the term runway is typically in the domain of
                                    > aeronautics and so one could ask, "Do you mean you need an airplane,
                                    > or do you need a firetruck, or do you need a baggage tractor?" But
                                    > maybe I am more familiar with the fashion industry and so I give you
                                    > a supermodel on roller blades.

                                    It has to pass the customer's tests. The example is amusing but
                                    perhaps too intentionally obtuse.

                                    > These scenarios are often presented in such a way so that there are
                                    > no communication paths available to the developer.

                                    Agile software development /requires/ a communication path to the
                                    developer. Everything I've been talking about so far assumes that
                                    context. On the other hand, without such a path, even if you're
                                    doing waterfall, you don't know whether they want a plane or a
                                    supermodel on wheels.

                                    > For example, suppose the "product" to be built is an airplane. The
                                    > scenario is as if the Project Manager divied up the tasks to
                                    > different development teams that can not talk to each other and that
                                    > the big picture is only given on a need to know basis. It is like the
                                    > PM told one group, "Build me something that can roll down a runway"
                                    > and he told another group to "build me something that can seat 8
                                    > passengers" and yet a third group he told "build me something that
                                    > can make things fly".

                                    Suppose the product is software. We're talking about software and
                                    how to do software. Airplanes are not software. The example won't
                                    fly.

                                    > There should be a big picture. You should know what is required. If
                                    > something comes up that you think may be needed then tell someone so
                                    > that it can be determined if something was overlooked. If it is not
                                    > an oversight then don't make it. Clearly simple design, regression
                                    > tests, properly defined domain, etc., can make the system easy to
                                    > extend and from my experience software is usually extended.

                                    Yes, software is usually extended. Yes, it's good to know where
                                    you're probably going to go.

                                    That emphatically does not imply that you should do extra work now
                                    to prepare for something that is going to happen later.

                                    > Many times I have heard, "xp doesn't mean check your brain at the
                                    > door". Making a big picture doesn't mean do BDUF. Understanding your
                                    > customer's requirements doesn't mean do BDUF. There is a planning
                                    > game to determing the product. Then this product (which is the big
                                    > picture) is divided up and releases and iterations are defined and
                                    > the product is delivered incrementally.

                                    > If you need a tractor, a plane, or a supermodel on roller blades, it
                                    > should be known by someone somewhere somehow.

                                    Sure. It's wonderful to know ...

                                    > I don't know, but it seems like I am missing something, or
                                    > misunderstanding something...

                                    Well, software isn't a tractor, a plane, or even a supermodel on
                                    skates. That could be part of the issue.

                                    The question I was originally answering had to do with whether, at a
                                    given iteration, the design should have in it things preparing for
                                    future iterations. My answer is that, other than being a good design
                                    for the features now in the system, there is no need to put stuff in
                                    for the future. Good design /is/ design for the future: modularity
                                    works.

                                    I don't know at all how to build an airplane by incremental
                                    development with refactoring. I do know a fair amount about how to
                                    build software that way, and I think I can be more helpful talking
                                    about that.

                                    Ron Jeffries
                                    www.XProgramming.com
                                    We cannot solve our problems with the same thinking we used when we created them.
                                    -- Albert Einstein
                                  • geoffrey_slinker
                                    (Sorry, but I made a very thought out reply and it seems to have disappeard...) Thanks for your time and your response. ... Yes! Good design /is/ design for
                                    Message 17 of 22 , Jun 9, 2006
                                    • 0 Attachment
                                      (Sorry, but I made a very thought out reply and it seems to have
                                      disappeard...)

                                      Thanks for your time and your response.

                                      >
                                      > The question I was originally answering had to do with whether, at a
                                      > given iteration, the design should have in it things preparing for
                                      > future iterations. My answer is that, other than being a good design
                                      > for the features now in the system, there is no need to put stuff in
                                      > for the future. Good design /is/ design for the future: modularity
                                      > works.

                                      Yes! Good design /is/ design for the future: modularity works.

                                      XP is the environment in which the developer can apply his skills.
                                      Good design is a skill that can be acquired through lots of study,
                                      experimentation, and application. XP is an excellent environment in
                                      which a developer is trusted and therefor allowed to practice his
                                      profession with excellence.

                                      Thanks,
                                      Geoff
                                    Your message has been successfully submitted and would be delivered to recipients shortly.