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

RE: [XP] Re: Honest inquiry

Expand Messages
  • Booch, Grady
    I agree. I just don t think mathematical modelling of an entire systems input and outputs before starting to code anything is a simplification. [egb ] I
    Message 1 of 15 , May 31, 2002
    • 0 Attachment
      I agree. I just don't think mathematical modelling of an entire
      systems input and outputs before starting to code anything is a
      simplification.
      [egb> ] I wasn't at the conference...is this your take on what Dave Parnas
      was saying?
    • twelve71
      ... Parnas ... That seemed (to me) to be *one* of the things he was saying. He figured any team would get a lot of mileage simply by codifying a set of all
      Message 2 of 15 , May 31, 2002
      • 0 Attachment
        --- In extremeprogramming@y..., "Booch, Grady" <egb@r...> wrote:
        > I agree. I just don't think mathematical modelling of an entire
        > systems input and outputs before starting to code anything is a
        > simplification.
        > [egb> ] I wasn't at the conference...is this your take on what Dave
        Parnas
        > was saying?

        That seemed (to me) to be *one* of the things he was saying. He
        figured any team would get a lot of mileage simply by codifying a set
        of all inputs and mapping them to outputs. I'll say it again to make
        sure I'm not misinterpreted... That's not *all* he was saying, just a
        part of it.

        My actual response (and Hill's) was "I can see how he did that in his
        queue example, and I can even see how he did it for the air-traf
        software he described working on in the 70s, but I have no idea how
        that could work for MS Word".

        Kent's comment at the closing speech was to ask if that was the sort
        of community we wanted to be: "If we can't see how to do it, it must
        be a stupid idea". I think Hill's response, after I told him what
        Kent had said in the speech was "I have no problem with new ideas, I
        just have a problem with folks bringing up the same old ideas".

        Alan
      • Ron Jeffries
        ... Parnas work is very interesting. I was very into it back then. I think it s something that one should practice and stay good at. That said, I don t know
        Message 3 of 15 , May 31, 2002
        • 0 Attachment
          Around Friday, May 31, 2002, 9:37:24 AM, twelve71 wrote:

          > --- In extremeprogramming@y..., "Booch, Grady" <egb@r...> wrote:
          >> I agree. I just don't think mathematical modelling of an entire
          >> systems input and outputs before starting to code anything is a
          >> simplification.
          >> [egb> ] I wasn't at the conference...is this your take on what Dave
          > Parnas
          >> was saying?

          > That seemed (to me) to be *one* of the things he was saying. He
          > figured any team would get a lot of mileage simply by codifying a set
          > of all inputs and mapping them to outputs. I'll say it again to make
          > sure I'm not misinterpreted... That's not *all* he was saying, just a
          > part of it.

          > My actual response (and Hill's) was "I can see how he did that in his
          > queue example, and I can even see how he did it for the air-traf
          > software he described working on in the 70s, but I have no idea how
          > that could work for MS Word".

          Parnas' work is very interesting. I was very into it back then. I
          think it's something that one should practice and stay good at.

          That said, I don't know how to do MS Word test first, for that matter.
          Both ways it's about breaking things down into pieces one's process
          can deal with, I guess.

          > Kent's comment at the closing speech was to ask if that was the sort
          > of community we wanted to be: "If we can't see how to do it, it must
          > be a stupid idea". I think Hill's response, after I told him what
          > Kent had said in the speech was "I have no problem with new ideas, I
          > just have a problem with folks bringing up the same old ideas".

          I get so tired of hearing that objection ...

          Ron Jeffries
          www.XProgramming.com
          Bang, bang, Jeffries' silver hammer came down upon their heads ...
        • Booch, Grady
          That said, I don t know how to do MS Word test first, for that matter. Both ways it s about breaking things down into pieces one s process can deal with, I
          Message 4 of 15 , May 31, 2002
          • 0 Attachment
            That said, I don't know how to do MS Word test first, for that matter.
            Both ways it's about breaking things down into pieces one's process
            can deal with, I guess.

            [egb> ] You can always simply cast it out to your customers, and let them
            test it. It's even better if you can get them to pay for the privilege of
            doing so...
          • Alan C Francis
            ... I also made that point. All the publicly available TfD examples are Money, or a STack or a very tiny Alarm Clock. Show me Eclipse.
            Message 5 of 15 , May 31, 2002
            • 0 Attachment
              > Parnas' work is very interesting. I was very into it back then. I
              > think it's something that one should practice and stay good at.
              >
              > That said, I don't know how to do MS Word test first, for that matter.
              > Both ways it's about breaking things down into pieces one's process
              > can deal with, I guess.

              I also made that point. All the publicly available TfD examples are Money,
              or a STack or a very tiny Alarm Clock. Show me Eclipse.
            • schroeder@surfree.com
              On Fri, 31 May 2002 16:23:50 +0100 Alan C ... It might be very interesting to see a log of test- driving a couple of new stories for a program like Eclipse or
              Message 6 of 15 , May 31, 2002
              • 0 Attachment
                On Fri, 31 May 2002 16:23:50 +0100 Alan C
                Francis <listsub@...> wrote:

                >> That said, I don't know how to do MS Word
                >> test first, for that matter.
                >> Both ways it's about breaking things down
                >> into pieces one's process
                >> can deal with, I guess.
                >
                > I also made that point. All the publicly
                > available TfD examples are Money,
                > or a STack or a very tiny Alarm Clock. Show
                > me Eclipse.

                It might be very interesting to see a log of test-
                driving a couple of new stories for a program
                like Eclipse or Word. A higher-level log of the
                stories implemented over time and how the
                objects' interactions changed might also be
                interesting.

                -- Ben Schroeder
                schroeder@...
              • Alan C Francis
                ... From: kentlbeck To: Sent: Friday, May 31, 2002 7:22 PM Subject: [XP] Re: Honest inquiry ... So it
                Message 7 of 15 , May 31, 2002
                • 0 Attachment
                  ---- Original Message -----
                  From: "kentlbeck" <kentbeck@...>
                  To: <extremeprogramming@yahoogroups.com>
                  Sent: Friday, May 31, 2002 7:22 PM
                  Subject: [XP] Re: Honest inquiry


                  > --- In extremeprogramming@y..., "twelve71" <acfrancis@a...> wrote:
                  > > I agree. I just don't think mathematical modelling of an entire
                  > > systems input and outputs before starting to code anything is a
                  > > simplification.
                  >
                  > So, don't do it all at once. We know we don't have to do that. Parnas
                  > doesn't. Replace source-code-based tests with tables of inputs and
                  > expected outputs written one row at a time. Won't work, you say? I'm
                  > not sure of that, because I haven't tried.

                  So it seems we're violently agreeing. You picked out and pointed out one
                  aspect of what I said because I picked out and pointed out one aspect of
                  what he said. I object to his manner and the approach he suggested as an
                  alternative to XP. Doesn't mean I don't think he has anything to say. I
                  have the Software Fundamentals book and I think there are many great papers
                  there. I think some of what he said in his talk was valid. I reserve my
                  right to disagree with some of it.

                  He recommended the tables because he seemed to feel that test-first tests
                  weren't sufficiently rigorous. I feel they *are*. He said my (actually
                  your) approach was inferior, I think his is just different (and less natural
                  for most programmers).

                  > Re: Parnas expressing his ideas offensively. Um, perhaps we as a
                  > community don't want to bringing up this particular point too
                  > strongly.

                  Fair point :-)

                  A.
                • Bill de hÓra
                  ... Hash: SHA1 ... Aren t Junit and mockobjects test first? Bill de hÓra ... Version: PGP 7.0.4
                  Message 8 of 15 , May 31, 2002
                  • 0 Attachment
                    -----BEGIN PGP SIGNED MESSAGE-----
                    Hash: SHA1


                    > -----Original Message-----
                    > From: Alan C Francis [mailto:listsub@...]
                    >
                    > I also made that point. All the publicly available TfD
                    > examples are Money,
                    > or a STack or a very tiny Alarm Clock. Show me Eclipse.

                    Aren't Junit and mockobjects test first?

                    Bill de hÓra


                    -----BEGIN PGP SIGNATURE-----
                    Version: PGP 7.0.4

                    iQA/AwUBPPfmf+aWiFwg2CH4EQJdjQCg4vneRJCXPlX+nbE+X8dRJc+/Q5oAmgNV
                    z457ZnA85KR2MgEBD6k7dzqM
                    =xUXH
                    -----END PGP SIGNATURE-----
                  • Alan C Francis
                    JUnit is, but it s still not a particularly large example... A. ... From: tranzpupy To: Sent:
                    Message 9 of 15 , May 31, 2002
                    • 0 Attachment
                      JUnit is, but it's still not a particularly large example...

                      A.
                      ----- Original Message -----
                      From: "tranzpupy" <tranzpupy@...>
                      To: <extremeprogramming@yahoogroups.com>
                      Sent: Friday, May 31, 2002 10:06 PM
                      Subject: [XP] Re: Honest inquiry


                      > --- In extremeprogramming@y..., "Laurent Bossavit" <laurent@b...>
                      > wrote:
                      > > > Thoughts?
                      > >
                      > > I agree with you that Coding Standard is the useless one. ;)
                      > >
                      > > I believe the customer/developer test brings value in the form of a
                      > > "pincer movement". I want to know how Emily and Geoffrey have been
                      > > getting the reported design quality.
                      > >
                      > > (Hi everyone. I had a great time at the conference. I'm back, but
                      > > you'll be hearing from me on this forum much less frequently -
                      > about
                      > > as much as from Kent is my guess.)
                      >
                      > And why, pray tell, is that?
                      >
                      > Kay
                      >
                      >
                      >
                      > To Post a message, send it to: extremeprogramming@...
                      >
                      > To Unsubscribe, send a blank message to:
                      extremeprogramming-unsubscribe@...
                      >
                      > ad-free courtesy of objectmentor.com
                      >
                      > Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
                      >
                      >
                    • Bill de hÓra
                      ... Hash: SHA1 ... Sure, it s probably what, under 5-6Kloc including tests? Keeping large an intuitive matter for now that probably has some correlation with
                      Message 10 of 15 , May 31, 2002
                      • 0 Attachment
                        -----BEGIN PGP SIGNED MESSAGE-----
                        Hash: SHA1


                        > -----Original Message-----
                        > From: Alan C Francis [mailto:listsub@...]
                        > Sent: 31 May 2002 22:39
                        > To: extremeprogramming@yahoogroups.com
                        > Subject: Re: [XP] Re: Honest inquiry
                        >
                        >
                        > JUnit is, but it's still not a particularly large example...

                        Sure, it's probably what, under 5-6Kloc including tests? Keeping
                        large an intuitive matter for now that probably has some
                        correlation with lines of code, what's the basis for resisting TDD
                        based on largeness so long as you're modularizing things? I'm
                        thinking if it can be done for 5Kloc it can be done for 10Kloc. Now
                        in my mind 30Kloc is an upper bound on a software module in a C
                        based language (and probably less than that for C itself), i.e., if
                        you have 50Kloc, look to modularize, you'll be glad you did. A
                        composition of twenty 30Kloc applications is likely to be easier to
                        control and maintain and reason about than a 500,000Kloc monoblock
                        java system (if not, why not?). though if you're building
                        multimillion line systems from large (100Kloc+) subsystems, TDD
                        scaling is possibly the least of your worries.

                        Essentially, I'm comfortable making the guess that TDD scales up to
                        the size of largest software module I'd like to build, which is
                        all I need it to scale to.

                        Bill de hÓra


                        -----BEGIN PGP SIGNATURE-----
                        Version: PGP 7.0.4

                        iQA/AwUBPPf3jeaWiFwg2CH4EQJpwQCfTdbZCxuAND+vzMNpWeKzY/jAZlEAniao
                        NKrbqr1+vCHIQ2JkhdvvMjpn
                        =7SEn
                        -----END PGP SIGNATURE-----
                      • Brian Button
                        Eclipse is pretty modular, so a new story right now can pretty easily be done with TDD. Its pieces are broken up into plugins, which are logically separate
                        Message 11 of 15 , May 31, 2002
                        • 0 Attachment
                          Eclipse is pretty modular, so a new story right now can pretty easily be
                          done with TDD. Its pieces are broken up into plugins, which are logically
                          separate units of functionality. As it stands now, each of these plugins has
                          hundreds of unit tests, and they've even built a way to run eclipse inside
                          eclipse so that you can run unit tests on plugins you are building. And the
                          developers of the refactoring plugin (org.eclipse.jdt.ui.refactoring IIRC)
                          are quite proud of the fact that they are practicing TDD as they add new
                          refactorings.

                          If anyone is interested, I'm building a plugin right now, and I'd be happy
                          to share some of what I've learned about developing plugins using TDD in
                          eclipse. All I want/need now is a pair...

                          bab
                          ---
                          Brian Button bbutton@...
                          Senior Consultant http://www.objectmentor.com
                          Object Mentor, Inc.

                          Extreme Programming in St. Louis http://groups.yahoo.com/group/xpstl
                          ----- Original Message -----
                          From: <schroeder@...>
                          To: <extremeprogramming@yahoogroups.com>
                          Sent: Friday, May 31, 2002 10:58 AM
                          Subject: Re: [XP] Re: Honest inquiry


                          > It might be very interesting to see a log of test-
                          > driving a couple of new stories for a program
                          > like Eclipse or Word. A higher-level log of the
                          > stories implemented over time and how the
                          > objects' interactions changed might also be
                          > interesting.
                        • kentlbeck
                          ... www.lifeware.ch is the largest TDD system I ve been involved in--250K lines of Smalltalk and about the same of tests, 5-10 people for four years. There are
                          Message 12 of 15 , May 31, 2002
                          • 0 Attachment
                            --- In extremeprogramming@y..., Bill de hÓra <dehora@e...> wrote:
                            >
                            > -----BEGIN PGP SIGNED MESSAGE-----
                            > Hash: SHA1
                            >
                            >
                            > > -----Original Message-----
                            > > From: Alan C Francis [mailto:listsub@t...]
                            > > Sent: 31 May 2002 22:39
                            > > To: extremeprogramming@y...
                            > > Subject: Re: [XP] Re: Honest inquiry
                            > >
                            > >
                            > > JUnit is, but it's still not a particularly large example...
                            >
                            > Sure, it's probably what, under 5-6Kloc including tests? Keeping
                            > large an intuitive matter for now that probably has some
                            > correlation with lines of code, what's the basis for resisting TDD
                            > based on largeness so long as you're modularizing things? I'm
                            > thinking if it can be done for 5Kloc it can be done for 10Kloc. Now
                            > in my mind 30Kloc is an upper bound on a software module in a C
                            > based language (and probably less than that for C itself), i.e., if
                            > you have 50Kloc, look to modularize, you'll be glad you did. A
                            > composition of twenty 30Kloc applications is likely to be easier to
                            > control and maintain and reason about than a 500,000Kloc monoblock
                            > java system (if not, why not?). though if you're building
                            > multimillion line systems from large (100Kloc+) subsystems, TDD
                            > scaling is possibly the least of your worries.
                            >
                            > Essentially, I'm comfortable making the guess that TDD scales up to
                            > the size of largest software module I'd like to build, which is
                            > all I need it to scale to.
                            >
                            > Bill de hÓra

                            www.lifeware.ch is the largest TDD system I've been involved in--250K
                            lines of Smalltalk and about the same of tests, 5-10 people for four
                            years. There are 4000 tests at all scales that run in 20 minutes. It
                            is a generic life insurance contract management system specialized
                            and in production for a dozen clients. The code is strongly
                            modularized (not that we can't do better).

                            Kent
                          • Dan Palanza
                            ... Hi Kent, I m interested in contract management, on a small scale, and so I m curious about applications done on a large scale: how much of this 250K lines
                            Message 13 of 15 , Jun 1, 2002
                            • 0 Attachment
                              Kent Beck wrote:

                              >www.lifeware.ch is the largest TDD system I've been involved in--250K
                              >lines of Smalltalk and about the same of tests, 5-10 people for four
                              >years. There are 4000 tests at all scales that run in 20 minutes. It
                              >is a generic life insurance contract management system specialized
                              >and in production for a dozen clients. The code is strongly
                              >modularized (not that we can't do better).

                              Hi Kent, I'm interested in contract management, on a small scale, and so
                              I'm curious about applications done on a large scale: how much of this 250K
                              lines of code is involved with double entry bookkeeping? For example, in
                              your own experience with contract management, is double entry bookkeeping
                              ever used as a resource to control the accuracy and completeness of a
                              contract's administration?

                              Dan


                              [Non-text portions of this message have been removed]
                            • kentlbeck
                              ... and so ... this 250K ... example, in ... bookkeeping ... a ... Absolutely. Every dollar/euro/whatever coming into or out of the system, or changing state
                              Message 14 of 15 , Jun 1, 2002
                              • 0 Attachment
                                --- In extremeprogramming@y..., Dan Palanza <dan@c...> wrote:
                                > Hi Kent, I'm interested in contract management, on a small scale,
                                and so
                                > I'm curious about applications done on a large scale: how much of
                                this 250K
                                > lines of code is involved with double entry bookkeeping? For
                                example, in
                                > your own experience with contract management, is double entry
                                bookkeeping
                                > ever used as a resource to control the accuracy and completeness of
                                a
                                > contract's administration?

                                Absolutely. Every dollar/euro/whatever coming into or out of the
                                system, or changing state in the system, is posted double entry
                                style. As for how many of the lines, I would guess it's a small
                                percentage like <5. However, critical advances in simplifying the
                                system were often preceded by new accounting insights (like
                                introducing pass-through accounts). When things go weird there is
                                nothing like an absolute complete history of what happened to all the
                                money to give you the raw material for good customer service.

                                Kent
                              • Benjamin Schroeder
                                ... Brian, I d be interested to know more about your experiences. Sorry for taking so long to reply to this post -- I was out over the weekend. Thanks, Ben
                                Message 15 of 15 , Jun 3, 2002
                                • 0 Attachment
                                  On Friday, May 31, 2002, at 07:47 PM, Brian Button wrote:

                                  > Eclipse is pretty modular, so a new story right now can pretty easily be
                                  > done with TDD. Its pieces are broken up into plugins, which are
                                  > logically
                                  > separate units of functionality. As it stands now, each of these
                                  > plugins has
                                  > hundreds of unit tests, and they've even built a way to run eclipse
                                  > inside
                                  > eclipse so that you can run unit tests on plugins you are building. And
                                  > the
                                  > developers of the refactoring plugin (org.eclipse.jdt.ui.refactoring
                                  > IIRC)
                                  > are quite proud of the fact that they are practicing TDD as they add new
                                  > refactorings.
                                  >
                                  > If anyone is interested, I'm building a plugin right now, and I'd be
                                  > happy
                                  > to share some of what I've learned about developing plugins using TDD in
                                  > eclipse. All I want/need now is a pair...

                                  Brian,

                                  I'd be interested to know more about your experiences. Sorry for taking
                                  so long to reply to this post -- I was out over the weekend.

                                  Thanks,
                                  Ben Schroeder
                                  schroeder@...
                                Your message has been successfully submitted and would be delivered to recipients shortly.