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

Goodbye to Eiffel?

Expand Messages
  • marvin_littlewood_426716
    Hello Everyone I have been wondering what to do with ISE Eiffel. I intend to wait for the upcoming improve quality (v6.3) release of Eiffel before deciding
    Message 1 of 19 , Nov 1, 2008
    • 0 Attachment
      Hello Everyone

      I have been wondering what to do with ISE Eiffel. I intend to wait
      for the upcoming "improve quality" (v6.3) release of Eiffel before
      deciding whether to continue using the language and EiffelStudio.
      What was a high quality tool when I started (v5.3) appears to have
      suffered in the face of a slew of new features.

      Perhaps my first issue is that the important ideas in Bertrand
      Meyer's book about Object Oriented Software Construction (which
      attracted me in the first place), such as concurrency and object
      persistence in databases remain to be implemented. A spec for SCOOP
      exists; isn't it time to incorporate it into the main stream?

      My second issue is that the Eiffel standard breaks too much existing
      code and removes features like indexing, which were a useful
      innovation in support of the generation of documentation. When I
      tried to compile for Void safety, a large number of bizarre errors
      turned up telling me that two features with the same text signature
      were in fact different but perhaps most annoying of all, I had to
      adopt cut and paste coding for create functions. This is not a
      useful "improvement" in a slow evolving standard and will not
      encourage me to adopt it.

      FWIW why replace LACE with the extra complexity of XML? Perhaps I am
      a luddite but LACE was easy to understand, use and occasionally edit.
      The interface to the arguably more capable XML almost always has to
      be through tools and is unpleasant to wade through when trying to
      understand configuration problems.

      My third issue is with EiffelStudio itself. In regard of the GUI, a
      useful view such as the class feature list is not properly updated
      after re-compiling - forcing a close and re-open of the editing
      window to get a partial and often incorrect listing of features.
      There are other annoyances like failing properly to remember the
      position of windows between invocations, forcing me to move sliders
      about to get the view I want. And what is the "favourites" view all
      about? I can't help wondering in these agile times, whether anyone
      spends long enough on a class to make it a "favourite". And why do I
      have to re-start EiffelStudio every time I want a clean compile?

      To me, the concept of having "Alt-F8" and "F7" is flawed: why have
      methods of compilation that allow broken code to compile (e.g. when
      feature availability changes in a library)? The user has to know
      stuff that should be the compiler's responsbility. I realise that a
      long term goal is to remove the need to press a compile button but I
      wonder about a compilation strategy that permits "surprises" to turn
      up when you "freeze" a project.

      My fourth issue is with the .NET implementation. In other tools such
      as Visual Studio, you can reference an assembly (e.g. Fit.dll) and
      the software sorts out the inclusion of relevant modules so you can
      get on with coding. In EiffelStudio you face having to sort out a
      VD71 error with no information about what is causing the problem in
      the fault message. Adding all the relevant assemblies that the
      original DLL references doesn't fix the problem and so what should be
      a five second job becomes hours of frustration. (BTW if software can
      enumerate these assemblies, e.g. AMEX, why can't EiffelStudio?)

      I would love to program in Eiffel. It's clarity and relative
      simplicity compared with languages like C/C++ (where I learnt my
      programming) make it very attractive. It offers one of the best
      expositions of what object-orientation is all about and integration
      with CASE and profiling tools make the environment potentially one of
      the best.

      However for me, EiffelStudio and the Eiffel Standard are starting to
      lose their way. I hope 6.3 will put us on the road to recovery but I
      am very concerned that this release will end up just being a holding
      operation before sliding further down a slippery slope of too many
      new features that don't quite work.

      Regards

      Mark
    • Peter Gummer
      ... Wow. My experience has been the exact opposite. I started with 5.4. It crashed so often I could hardly believe it. I have no such problems these days: the
      Message 2 of 19 , Nov 1, 2008
      • 0 Attachment
        marvin_littlewood_426716 wrote:
        > What was a high quality tool when I started (v5.3) appears to have
        > suffered in the face of a slew of new features.
        >

        Wow. My experience has been the exact opposite. I started with 5.4. It
        crashed so often I could hardly believe it. I have no such problems
        these days: the quality has improved enormously.

        > My second issue is that the Eiffel standard breaks too much existing
        > code and removes features like indexing ...

        The standard hasn't removed "indexing", it's renamed it to "note".

        And what code is broken? I've been amazed at how I've had to change only
        a handful of lines of code due to the changes of the last few releases.

        > ... When I tried to compile for Void safety ...

        You're brave to have attempted that! So you tried an interesting
        experiment that didn't work ... just go back to the old way.

        I expect the transition to void safety will be gradual. We've been using
        attachment markers, to get the benefit of compile-time checking.


        > My third issue is with EiffelStudio itself. In regard of the GUI, a
        > useful view such as the class feature list is not properly updated
        > after re-compiling - forcing a close and re-open of the editing
        > window to get a partial and often incorrect listing of features.
        > There are other annoyances like failing properly to remember the
        > position of windows between invocations, forcing me to move sliders
        > about to get the view I want. ... And why do I
        > have to re-start EiffelStudio every time I want a clean compile?
        >

        I don't have any of these problems. They always work for me.

        It's strange that we have such different experiences of the last few
        releases. What kinds of projects are you building?

        - Peter Gummer
      • azador1606@yahoo.com
        I agree with Marvin. I have dropped ISE like the lemon it is. But the language is still fundamentally sound and has great potential. Old versions still work.
        Message 3 of 19 , Nov 1, 2008
        • 0 Attachment
          I agree with Marvin. I have dropped ISE like the lemon it is.

          But the language is still fundamentally sound and has great potential.

          Old versions still work. Of course a half decent .NET Eiffel compiler would make my day. Sigh!
          Sent via BlackBerry® from Vodafone

          -----Original Message-----
          From: Peter Gummer <p-gummer@...>

          Date: Sun, 02 Nov 2008 00:13:10
          To: <eiffel_software@yahoogroups.com>
          Subject: Re: [eiffel_software] Goodbye to Eiffel?


          marvin_littlewood_426716 wrote:
          > What was a high quality tool when I started (v5.3) appears to have
          > suffered in the face of a slew of new features.
          >

          Wow. My experience has been the exact opposite. I started with 5.4. It
          crashed so often I could hardly believe it. I have no such problems
          these days: the quality has improved enormously.

          > My second issue is that the Eiffel standard breaks too much existing
          > code and removes features like indexing ...

          The standard hasn't removed "indexing", it's renamed it to "note".

          And what code is broken? I've been amazed at how I've had to change only
          a handful of lines of code due to the changes of the last few releases.

          > ... When I tried to compile for Void safety ...

          You're brave to have attempted that! So you tried an interesting
          experiment that didn't work ... just go back to the old way.

          I expect the transition to void safety will be gradual. We've been using
          attachment markers, to get the benefit of compile-time checking.


          > My third issue is with EiffelStudio itself. In regard of the GUI, a
          > useful view such as the class feature list is not properly updated
          > after re-compiling - forcing a close and re-open of the editing
          > window to get a partial and often incorrect listing of features.
          > There are other annoyances like failing properly to remember the
          > position of windows between invocations, forcing me to move sliders
          > about to get the view I want. ... And why do I
          > have to re-start EiffelStudio every time I want a clean compile?
          >

          I don't have any of these problems. They always work for me.

          It's strange that we have such different experiences of the last few
          releases. What kinds of projects are you building?

          - Peter Gummer





          [Non-text portions of this message have been removed]
        • Helmut Brandl
          ... Can you give me a reference to the spec of SCOOP. I am not aware of a specification, but I would highly appreciate any information. Regards Helmut The
          Message 4 of 19 , Nov 2, 2008
          • 0 Attachment
            marvin_littlewood_426716 wrote:
            > Hello Everyone
            >
            > I have been wondering what to do with ISE Eiffel. I intend to wait
            > for the upcoming "improve quality" (v6.3) release of Eiffel before
            > deciding whether to continue using the language and EiffelStudio.
            > What was a high quality tool when I started (v5.3) appears to have
            > suffered in the face of a slew of new features.
            >
            > Perhaps my first issue is that the important ideas in Bertrand
            > Meyer's book about Object Oriented Software Construction (which
            > attracted me in the first place), such as concurrency and object
            > persistence in databases remain to be implemented. A spec for SCOOP
            > exists; isn't it time to incorporate it into the main stream?
            >
            >
            Can you give me a reference to the spec of SCOOP. I am not aware of a
            specification, but I would highly appreciate any information.

            Regards
            Helmut

            The Eiffel Compiler: http://tecomp.sourceforge.net
            http://www.sourceforge.net/projects/tecomp
          • javier hector
            You can see http://scoop.origo.ethz.ch/ ________________________________ De: Helmut Brandl Para: eiffel_software@yahoogroups.com
            Message 5 of 19 , Nov 2, 2008
            • 0 Attachment
              You can see http://scoop.origo.ethz.ch/



              ________________________________
              De: Helmut Brandl <helmut.brandl@...>
              Para: eiffel_software@yahoogroups.com
              Enviado: domingo 2 de noviembre de 2008, 15:02:51
              Asunto: Re: [eiffel_software] Goodbye to Eiffel?


              marvin_littlewood_ 426716 wrote:
              > Hello Everyone
              >
              > I have been wondering what to do with ISE Eiffel. I intend to wait
              > for the upcoming "improve quality" (v6.3) release of Eiffel before
              > deciding whether to continue using the language and EiffelStudio.
              > What was a high quality tool when I started (v5.3) appears to have
              > suffered in the face of a slew of new features.
              >
              > Perhaps my first issue is that the important ideas in Bertrand
              > Meyer's book about Object Oriented Software Construction (which
              > attracted me in the first place), such as concurrency and object
              > persistence in databases remain to be implemented. A spec for SCOOP
              > exists; isn't it time to incorporate it into the main stream?
              >
              >
              Can you give me a reference to the spec of SCOOP. I am not aware of a
              specification, but I would highly appreciate any information.

              Regards
              Helmut

              The Eiffel Compiler: http://tecomp. sourceforge. net
              http://www.sourcefo rge.net/projects /tecomp



              Yahoo! Cocina
              Recetas prácticas y comida saludable
              http://ar.mujer.yahoo.com/cocina/

              [Non-text portions of this message have been removed]
            • marvin_littlewood_426716
              Hello again! Thank you for your thoughtful replies: Roger s impassioned plea, Azador s rejection of ISE in favour of GOBO and Peter s different experience with
              Message 6 of 19 , Nov 2, 2008
              • 0 Attachment
                Hello again!

                Thank you for your thoughtful replies: Roger's impassioned plea,
                Azador's rejection of ISE in favour of GOBO and Peter's different
                experience with ISE's GUI and tools.

                Let me be clear: I buy into Eiffel big time. I don't want to stop
                using the language or the ISE tools; they pack a lot of punch and do
                an awful lot right. However releases to improve quality worry me.
                Specifically they raise the question where has the quality control
                been beforehand? I raised my issues previously in order to prevent a
                slide down from the high standards that have been quite well met, so
                far.

                My usual way (nowadays) of getting into a language is to build a unit
                test tool. The project that caused all my problems was a port of
                jUnit 3.8 to Eiffel. Every so often after an F7 compile it would fail
                its tests with a segmentation violation. A freeze generally solved
                the problem. This sort of intermittent failure, which disappeared as
                soon as I tried to investigate it, is the sort of dead end that leads
                me to lose a little confidence.

                I wondered if there was an uninitialised pointer. I tried compiling
                using the Void safety settings to see if that would help. At that
                point I got about 100 error messages. 90% were easy to fix. 10% were
                truly mystifying and I have posted about those and sent a bug report.
                Reverting to the old settings got rid of the error messages but did
                not fix the segmentation violation.

                On seeing that Void safety effectively meant duplicating code in
                creation functions, I became even more concerned. Surely a purist's
                language like Eiffel cannot be advocating cut and paste coding? OK so
                there is a solution: write one creation function and force users to
                use post-creation setup functions, but that seems inelegant.

                Out of interest, I tried compiling for standard Eiffel. Finding the
                issue about Indexing and a comment about "Note" not yet being
                available, I was a little perplexed. Having to modify all my classes
                for no perceivable benefit rather put me off proceeding.

                The idea that it is brave to try to use reference types and the
                standard seems a little strange... Surely the biggest contribution
                anyone can make is to adopt these things as quickly as possible and
                submit bug reports. Experience in C++ suggests that having a standard
                is a benefit to all concerned.

                As regards .NET, it is disappointing that a problem in 5.7 remains in
                6.2. Happily the route to a fix was posted in this newsgroup...

                The GUI stuff is a minor annoyance and appears to work for others so
                perhaps it is something on my system. I'll happily see what happens
                next version.

                So thanks once again. I am temporarily reassured and don't plan to
                throw the baby out with the bathwater! Yet!!

                Regards

                Mark
              • Thomas Beale
                As for Peter Gummer, I certainly couldn t agree with this experience either. I have used Eiffel since its inception, and early implementations were unreliable,
                Message 7 of 19 , Nov 2, 2008
                • 0 Attachment
                  As for Peter Gummer, I certainly couldn't agree with this experience
                  either. I have used Eiffel since its inception, and early
                  implementations were unreliable, but that situation changed some years
                  ago. The current environment is quite a pleasure to use. I would not
                  expect that void-safe code could just be achieved overnight; it is a
                  profound design change in the language (for the better) and code built
                  on pre-void safe thinking will just be by its nature different.

                  I think that the Eiffel Software team are doing a very good job with the
                  tool. I don't agree with absolutely everything being done in the
                  language and undoubtedly there is some level of bugs in the tools. But
                  not more than Visual Studio or Eclipse. The main thing is that internal
                  consistency and a solid mathematical basis for the language be
                  maintained, and I think this remains the case.

                  - thomas beale


                  marvin_littlewood_426716 wrote:
                  >
                  > Hello Everyone
                  >
                  > I have been wondering what to do with ISE Eiffel. I intend to wait
                  > for the upcoming "improve quality" (v6.3) release of Eiffel before
                  > deciding whether to continue using the language and EiffelStudio.
                  > What was a high quality tool when I started (v5.3) appears to have
                  > suffered in the face of a slew of new features.
                  >
                  > Perhaps my first issue is that the important ideas in Bertrand
                  > Meyer's book about Object Oriented Software Construction (which
                  > attracted me in the first place), such as concurrency and object
                  > persistence in databases remain to be implemented. A spec for SCOOP
                  > exists; isn't it time to incorporate it into the main stream?
                  >
                  > My second issue is that the Eiffel standard breaks too much existing
                  > code and removes features like indexing, which were a useful
                  > innovation in support of the generation of documentation. When I
                  > tried to compile for Void safety, a large number of bizarre errors
                  > turned up telling me that two features with the same text signature
                  > were in fact different but perhaps most annoying of all, I had to
                  > adopt cut and paste coding for create functions. This is not a
                  > useful "improvement" in a slow evolving standard and will not
                  > encourage me to adopt it.
                  >
                  > FWIW why replace LACE with the extra complexity of XML? Perhaps I am
                  > a luddite but LACE was easy to understand, use and occasionally edit.
                  > The interface to the arguably more capable XML almost always has to
                  > be through tools and is unpleasant to wade through when trying to
                  > understand configuration problems.
                  >
                  > My third issue is with EiffelStudio itself. In regard of the GUI, a
                  > useful view such as the class feature list is not properly updated
                  > after re-compiling - forcing a close and re-open of the editing
                  > window to get a partial and often incorrect listing of features.
                  > There are other annoyances like failing properly to remember the
                  > position of windows between invocations, forcing me to move sliders
                  > about to get the view I want. And what is the "favourites" view all
                  > about? I can't help wondering in these agile times, whether anyone
                  > spends long enough on a class to make it a "favourite". And why do I
                  > have to re-start EiffelStudio every time I want a clean compile?
                  >
                  > To me, the concept of having "Alt-F8" and "F7" is flawed: why have
                  > methods of compilation that allow broken code to compile (e.g. when
                  > feature availability changes in a library)? The user has to know
                  > stuff that should be the compiler's responsbility. I realise that a
                  > long term goal is to remove the need to press a compile button but I
                  > wonder about a compilation strategy that permits "surprises" to turn
                  > up when you "freeze" a project.
                  >
                  > My fourth issue is with the .NET implementation. In other tools such
                  > as Visual Studio, you can reference an assembly (e.g. Fit.dll) and
                  > the software sorts out the inclusion of relevant modules so you can
                  > get on with coding. In EiffelStudio you face having to sort out a
                  > VD71 error with no information about what is causing the problem in
                  > the fault message. Adding all the relevant assemblies that the
                  > original DLL references doesn't fix the problem and so what should be
                  > a five second job becomes hours of frustration. (BTW if software can
                  > enumerate these assemblies, e.g. AMEX, why can't EiffelStudio?)
                  >
                  > I would love to program in Eiffel. It's clarity and relative
                  > simplicity compared with languages like C/C++ (where I learnt my
                  > programming) make it very attractive. It offers one of the best
                  > expositions of what object-orientation is all about and integration
                  > with CASE and profiling tools make the environment potentially one of
                  > the best.
                  >
                  > However for me, EiffelStudio and the Eiffel Standard are starting to
                  > lose their way. I hope 6.3 will put us on the road to recovery but I
                  > am very concerned that this release will end up just being a holding
                  > operation before sliding further down a slippery slope of too many
                  > new features that don't quite work.
                  >
                  > Regards
                  >
                  > Mark
                  >
                  >


                  --
                  *Thomas Beale
                  Chief Technology Officer, Ocean Informatics
                  <http://www.oceaninformatics.com/>*

                  Chair Architectural Review Board, /open/EHR Foundation
                  <http://www.openehr.org/>
                  Honorary Research Fellow, University College London
                  <http://www.chime.ucl.ac.uk/>


                  *
                  *
                • Helmut Brandl
                  Thank you for the link. It is an interesting contribution. But I have not found any specification of SCOOP on that site. I was wondering whether Mark has some
                  Message 8 of 19 , Nov 2, 2008
                  • 0 Attachment
                    Thank you for the link.

                    It is an interesting contribution. But I have not found any
                    specification of SCOOP on that site. I was wondering whether Mark has
                    some other information, because he wrote of a "spec" for SCOOP.

                    Helmut
                    The Eiffel Compiler: http://tecomp.sourceforge.net
                    http://www.sourceforge.net/projects/tecomp

                    javier hector wrote:
                    > You can see http://scoop.origo.ethz.ch/
                    >
                    >
                    >
                    > ________________________________
                    > De: Helmut Brandl <helmut.brandl@...>
                    > Para: eiffel_software@yahoogroups.com
                    > Enviado: domingo 2 de noviembre de 2008, 15:02:51
                    > Asunto: Re: [eiffel_software] Goodbye to Eiffel?
                    >
                    >
                    > marvin_littlewood_ 426716 wrote:
                    >
                    >> Hello Everyone
                    >>
                    >> I have been wondering what to do with ISE Eiffel. I intend to wait
                    >> for the upcoming "improve quality" (v6.3) release of Eiffel before
                    >> deciding whether to continue using the language and EiffelStudio.
                    >> What was a high quality tool when I started (v5.3) appears to have
                    >> suffered in the face of a slew of new features.
                    >>
                    >> Perhaps my first issue is that the important ideas in Bertrand
                    >> Meyer's book about Object Oriented Software Construction (which
                    >> attracted me in the first place), such as concurrency and object
                    >> persistence in databases remain to be implemented. A spec for SCOOP
                    >> exists; isn't it time to incorporate it into the main stream?
                    >>
                    >>
                    >>
                    > Can you give me a reference to the spec of SCOOP. I am not aware of a
                    > specification, but I would highly appreciate any information.
                    >
                    > Regards
                    > Helmut
                    >
                    > The Eiffel Compiler: http://tecomp.sourceforge.net
                    > http://www.sourceforge.net/projects/tecomp
                    >
                    >
                    >
                  • Helmut Brandl
                    ... Mark, is that the issue that with the current version of the ECMA specification it is not allowed to delegate initialization of attributes to features that
                    Message 9 of 19 , Nov 2, 2008
                    • 0 Attachment
                      marvin_littlewood_426716 wrote:
                      > Hello again!
                      >
                      > Thank you for your thoughtful replies: Roger's impassioned plea,
                      > Azador's rejection of ISE in favour of GOBO and Peter's different
                      > experience with ISE's GUI and tools.
                      >
                      >
                      ....
                      > On seeing that Void safety effectively meant duplicating code in
                      > creation functions, I became even more concerned. Surely a purist's
                      > language like Eiffel cannot be advocating cut and paste coding? OK so
                      > there is a solution: write one creation function and force users to
                      > use post-creation setup functions, but that seems inelegant.
                      >
                      Mark,

                      is that the issue that with the current version of the ECMA
                      specification it is not allowed to delegate initialization of attributes
                      to features that the creation procedure might call?

                      If you mean that, I hope that the ECMA committee will change that
                      restriction. It should be possible to do the initialization of
                      attributes also in routines called by the creation procedures. It is not
                      difficult to implement, so a good compiler which implements void safety
                      will implement it anyhow.

                      Post creation setup functions may be a workaround but not a solution to
                      the problem.

                      Maybe the ECMA members on this list can publish their opinion about that
                      restriction in the ECMA spec. The users of Eiffel, especially those who
                      consider void safety as an improvement, would appreciate it. Especially
                      writers of Eiffel compilers like me need some indication of the
                      direction they want to go.

                      In order to be successful, Eiffel needs a rigid language specification
                      and some compilers which are committed to implement that language.

                      There is also the restriction in the ECMA spec that it disallows
                      initialization of attributes in a different manner in different branches
                      of "if" statements. But initialization is safe as long as it is done in
                      all possible branches including a mandatory "else" branch. As far as I
                      know, the ISE Eiffel compiler has already implemented some
                      "improvements", but I would prefer a written specification.

                      Eiffel usually puts the burden on the writer of the compiler an not on
                      the user. Regarding void safety the ECMA committee did the opposite.


                      > Experience in C++ suggests that having a standard
                      > is a benefit to all concerned.
                      >
                      As long as the standardization committee really works .... Maybe we hear
                      something before or after the next ECMA meeting which seems to be
                      scheduled for January 2009.


                      Helmut

                      The Eiffel Compiler: http://tecomp.sourceforge.net
                      http://www.sourceforge.net/projects/tecomp
                    • azador1606@yahoo.com
                      I should clarify that I only reject the ISE implementation of .NET Eiffel. Their IDE for Eiffel is a fine product I although I ve never been able to compile it
                      Message 10 of 19 , Nov 2, 2008
                      • 0 Attachment
                        I should clarify that I only reject the ISE implementation of .NET Eiffel. Their IDE for Eiffel is a fine product I although I've never been able to compile it on 64 bit Linux.

                        Sent via BlackBerry® from Vodafone

                        -----Original Message-----
                        From: "marvin_littlewood_426716" <marvin_littlewood_426716@...>

                        Date: Sun, 02 Nov 2008 22:43:18
                        To: <eiffel_software@yahoogroups.com>
                        Subject: [eiffel_software] Re: Goodbye to Eiffel?


                        Hello again!

                        Thank you for your thoughtful replies: Roger's impassioned plea,
                        Azador's rejection of ISE in favour of GOBO and Peter's different
                        experience with ISE's GUI and tools.

                        Let me be clear: I buy into Eiffel big time. I don't want to stop
                        using the language or the ISE tools; they pack a lot of punch and do
                        an awful lot right. However releases to improve quality worry me.
                        Specifically they raise the question where has the quality control
                        been beforehand? I raised my issues previously in order to prevent a
                        slide down from the high standards that have been quite well met, so
                        far.

                        My usual way (nowadays) of getting into a language is to build a unit
                        test tool. The project that caused all my problems was a port of
                        jUnit 3.8 to Eiffel. Every so often after an F7 compile it would fail
                        its tests with a segmentation violation. A freeze generally solved
                        the problem. This sort of intermittent failure, which disappeared as
                        soon as I tried to investigate it, is the sort of dead end that leads
                        me to lose a little confidence.

                        I wondered if there was an uninitialised pointer. I tried compiling
                        using the Void safety settings to see if that would help. At that
                        point I got about 100 error messages. 90% were easy to fix. 10% were
                        truly mystifying and I have posted about those and sent a bug report.
                        Reverting to the old settings got rid of the error messages but did
                        not fix the segmentation violation.

                        On seeing that Void safety effectively meant duplicating code in
                        creation functions, I became even more concerned. Surely a purist's
                        language like Eiffel cannot be advocating cut and paste coding? OK so
                        there is a solution: write one creation function and force users to
                        use post-creation setup functions, but that seems inelegant.

                        Out of interest, I tried compiling for standard Eiffel. Finding the
                        issue about Indexing and a comment about "Note" not yet being
                        available, I was a little perplexed. Having to modify all my classes
                        for no perceivable benefit rather put me off proceeding.

                        The idea that it is brave to try to use reference types and the
                        standard seems a little strange... Surely the biggest contribution
                        anyone can make is to adopt these things as quickly as possible and
                        submit bug reports. Experience in C++ suggests that having a standard
                        is a benefit to all concerned.

                        As regards .NET, it is disappointing that a problem in 5.7 remains in
                        6.2. Happily the route to a fix was posted in this newsgroup...

                        The GUI stuff is a minor annoyance and appears to work for others so
                        perhaps it is something on my system. I'll happily see what happens
                        next version.

                        So thanks once again. I am temporarily reassured and don't plan to
                        throw the baby out with the bathwater! Yet!!

                        Regards

                        Mark




                        [Non-text portions of this message have been removed]
                      • marvin_littlewood_426716
                        Helmut In answer to the issue of a spec, I was referring to this thesis which seemed to me to provide enough material thoroughly to understand the issues and
                        Message 11 of 19 , Nov 4, 2008
                        • 0 Attachment
                          Helmut

                          In answer to the issue of a spec, I was referring to this thesis
                          which seemed to me to provide enough material thoroughly to
                          understand the issues and how one might approach solving them:

                          http://cs.anu.edu.au/~Richard.Walker/eiffel/scoop/mc-thesis.pdf

                          As regards what you say below...

                          --- In eiffel_software@yahoogroups.com, Helmut Brandl
                          <helmut.brandl@...> wrote:
                          > > On seeing that Void safety effectively meant duplicating code in
                          > > creation functions, I became even more concerned. Surely a
                          purist's
                          > > language like Eiffel cannot be advocating cut and paste coding? >
                          is that the issue that with the current version of the ECMA
                          > specification it is not allowed to delegate initialization of
                          attributes
                          > to features that the creation procedure might call?
                          >
                          > If you mean that, I hope that the ECMA committee will change that
                          > restriction. It should be possible to do the initialization of
                          > attributes also in routines called by the creation procedures. It
                          is not
                          > difficult to implement, so a good compiler which implements void
                          safety
                          > will implement it anyhow.
                          >
                          > Post creation setup functions may be a workaround but not a
                          solution to
                          > the problem.
                          >
                          > Maybe the ECMA members on this list can publish their opinion about
                          that
                          > restriction in the ECMA spec. The users of Eiffel, especially
                          those who
                          > consider void safety as an improvement, would appreciate it.

                          My approach was from a practical point of view... I had a class which
                          had multiple create functions and modifying it to meet this
                          restriction resulted in cut and paste coding - something that I am
                          dogmatic about avoiding as I have seen it kill perfectly good
                          software. Putting the rules into a relatively slow moving standard,
                          seems to me to be a good way of delaying the standard's adoption.

                          There is an analogous argument to be made in regard of the removal
                          of "indexing" and its replacement with "note".

                          I should add one further point, I am not sure the discussion about
                          namespaces is over either. The standard suggests that it will be easy
                          for users to change a class name if the tool identifies a conflict.
                          This seems to me to disadvantage library writers. Also it puts users
                          of libraries in a difficult position if a conflict were to occur
                          between two read-only libraries. However I am perfectly willing to
                          accept that my conception of the issue may be mistaken.

                          FWIW I have adopted a strategy of prefixing my classes with the names
                          of the cluster within which they reside but this leads to long class
                          names, which isn't always ideal. Here is one way in which the newer
                          tools really are a big help.

                          Regards

                          Mark
                        • marvin_littlewood_426716
                          Hello ... It turns out that the VD71 problem on adding fit.dll is in part an issue of permissions. Running the compiler as an administrator resulted in a
                          Message 12 of 19 , Nov 4, 2008
                          • 0 Attachment
                            Hello

                            --- In eiffel_software@yahoogroups.com, "marvin_littlewood_426716"
                            <marvin_littlewood_426716@...> wrote:
                            >
                            > As regards .NET, it is disappointing that a problem in 5.7 remains in
                            > 6.2. Happily the route to a fix was posted in this newsgroup...
                            >

                            It turns out that the VD71 problem on adding fit.dll is in part an
                            issue of permissions. Running the compiler as an "administrator"
                            resulted in a compile. I normally operate as a "user"; hence the fault.

                            Regards

                            Mark
                          • Peter Gummer
                            ... Perhaps the fact that my experience of EiffelStudio 6.2 has been more positive than yours, Mark, can be attributed to the fact that I m still using Windows
                            Message 13 of 19 , Nov 4, 2008
                            • 0 Attachment
                              marvin_littlewood_426716 wrote:
                              > It turns out that the VD71 problem on adding fit.dll is in part an
                              > issue of permissions. Running the compiler as an "administrator"
                              > resulted in a compile. I normally operate as a "user"; hence the fault.
                              >

                              Perhaps the fact that my experience of EiffelStudio 6.2 has been more
                              positive than yours, Mark, can be attributed to the fact that I'm still
                              using Windows XP and Visual Studio 2005.

                              - Peter
                            • Alexander Kogtenkov
                              ... This is addresses in the forthcoming ISO/ECMA standard as well as in EiffelStudio release. ... At the moment EiffelStudio allows initialization of locals
                              Message 14 of 19 , Nov 4, 2008
                              • 0 Attachment
                                Helmut Brandl wrote:

                                > is that the issue that with the current version of the ECMA
                                > specification it is not allowed to delegate initialization of attributes
                                > to features that the creation procedure might call?

                                This is addresses in the forthcoming ISO/ECMA standard as well as
                                in EiffelStudio release.

                                > There is also the restriction in the ECMA spec that it disallows
                                > initialization of attributes in a different manner in different branches
                                > of "if" statements. But initialization is safe as long as it is done in
                                > all possible branches including a mandatory "else" branch. As far as I
                                > know, the ISE Eiffel compiler has already implemented some
                                > "improvements", but I would prefer a written specification.

                                At the moment EiffelStudio allows initialization of locals in different
                                branches of a conditional instruction. So, it can be used as a workaround
                                to initialize attributes in this way if required until the similar rules are
                                adopted for attributes as well.

                                Regards,
                                Alexander Kogtenkov
                              • Helmut Brandl
                                Hi Alexander, thank you for your response. Some remarks/questions below. Helmut The Eiffel Compiler: http://tecomp.sourceforge.net
                                Message 15 of 19 , Nov 4, 2008
                                • 0 Attachment
                                  Hi Alexander,

                                  thank you for your response. Some remarks/questions below.

                                  Helmut

                                  The Eiffel Compiler: http://tecomp.sourceforge.net
                                  http://www.sourceforge.net/projects/tecomp

                                  Alexander Kogtenkov wrote:
                                  > Helmut Brandl wrote:
                                  >
                                  >
                                  >> is that the issue that with the current version of the ECMA
                                  >> specification it is not allowed to delegate initialization of attributes
                                  >> to features that the creation procedure might call?
                                  >>
                                  >
                                  > This is addresses in the forthcoming ISO/ECMA standard as well as
                                  > in EiffelStudio release.
                                  >
                                  Could you be more specific? Is there an agreement to address that issue
                                  or is there already an agreement on how to address it? Manu indicated to
                                  me that I cannot expect a new release of the standard before 2010. So I
                                  read your answer like "Just wait one year and you will see".

                                  If there is already a direction known in the committee, would you be so
                                  kind to disclose it.
                                  >
                                  >> There is also the restriction in the ECMA spec that it disallows
                                  >> initialization of attributes in a different manner in different branches
                                  >> of "if" statements. But initialization is safe as long as it is done in
                                  >> all possible branches including a mandatory "else" branch. As far as I
                                  >> know, the ISE Eiffel compiler has already implemented some
                                  >> "improvements", but I would prefer a written specification.
                                  >>
                                  >
                                  > At the moment EiffelStudio allows initialization of locals in different
                                  > branches of a conditional instruction. So, it can be used as a workaround
                                  > to initialize attributes in this way if required until the similar rules are
                                  > adopted for attributes as well.
                                  >
                                  I have read about the possibility to initialize local variables in
                                  different branches of a conditional at the origo site. But in the
                                  discussion there were strong opinions not to allow that for attributes,
                                  because unlike locals, attributes can be changed by called features
                                  (even by innocent looking queries). Since the discussion I have read is
                                  old, I don't know, if the opinions have changed.
                                • azador1606@yahoo.com
                                  Whereas I use 32 bit VS2008 or 64 bit Linux both of which are no end of problems (for ISE). Sent via BlackBerry® from Vodafone ... From: Peter Gummer
                                  Message 16 of 19 , Nov 4, 2008
                                  • 0 Attachment
                                    Whereas I use 32 bit VS2008 or 64 bit Linux both of which are no end of problems (for ISE).
                                    Sent via BlackBerry® from Vodafone

                                    -----Original Message-----
                                    From: Peter Gummer <p-gummer@...>

                                    Date: Wed, 05 Nov 2008 00:32:21
                                    To: <eiffel_software@yahoogroups.com>
                                    Subject: Re: [eiffel_software] Re: Goodbye to Eiffel?


                                    marvin_littlewood_426716 wrote:
                                    > It turns out that the VD71 problem on adding fit.dll is in part an
                                    > issue of permissions. Running the compiler as an "administrator"
                                    > resulted in a compile. I normally operate as a "user"; hence the fault.
                                    >

                                    Perhaps the fact that my experience of EiffelStudio 6.2 has been more
                                    positive than yours, Mark, can be attributed to the fact that I'm still
                                    using Windows XP and Visual Studio 2005.

                                    - Peter




                                    [Non-text portions of this message have been removed]
                                  • marvin_littlewood_426716
                                    Hello! ... of problems (for ISE). ... still ... Actually I use Win XP sp3 and VS2008 on a 32 bit platform... In regard of my EiffelStudio GUI problems, the
                                    Message 17 of 19 , Nov 6, 2008
                                    • 0 Attachment
                                      Hello!

                                      --- In eiffel_software@yahoogroups.com, azador1606@... wrote:
                                      >
                                      > Whereas I use 32 bit VS2008 or 64 bit Linux both of which are no end
                                      of problems (for ISE).
                                      > Sent via BlackBerry® from Vodafone
                                      >
                                      > -----Original Message-----
                                      > From: Peter Gummer <p-gummer@...>
                                      >
                                      >
                                      > Perhaps the fact that my experience of EiffelStudio 6.2 has been more
                                      > positive than yours, Mark, can be attributed to the fact that I'm
                                      still
                                      > using Windows XP and Visual Studio 2005.
                                      >

                                      Actually I use Win XP sp3 and VS2008 on a 32 bit platform...

                                      In regard of my EiffelStudio GUI problems, the saved window size not
                                      being restored may be related to my screen resolution: 1680 by 1050.
                                      The issue of the "Features" tab failing to update after an F7 compile,
                                      unless I re-open the editor window is easily replicated on my system
                                      but may be unique to my setup. As I said previously these are minor
                                      annoyances rather than major problems.

                                      Out of habit, I tend to operate in User mode rather than as an
                                      Administrator. This opens up a question: should the Metadata consumer
                                      tool work in user mode? Or is there an unknown dependency on running as
                                      an administrator?

                                      Thanks

                                      Kind regards

                                      Mark
                                    • azador1606@yahoo.com
                                      Any metadata consumer must be able to consume in user mode. However, that doesn t mean you will have complete access at runtime necessarily - but it would be
                                      Message 18 of 19 , Nov 6, 2008
                                      • 0 Attachment
                                        Any metadata consumer must be able to consume in user mode.
                                        However, that doesn't mean you will have complete access at runtime necessarily - but it would be exceptional not to.
                                        Sent via BlackBerry® from Vodafone

                                        -----Original Message-----
                                        From: "marvin_littlewood_426716" <marvin_littlewood_426716@...>

                                        Date: Thu, 06 Nov 2008 12:18:21
                                        To: <eiffel_software@yahoogroups.com>
                                        Subject: [eiffel_software] Re: Goodbye to Eiffel?


                                        Hello!

                                        --- In eiffel_software@yahoogroups.com, azador1606@... wrote:
                                        >
                                        > Whereas I use 32 bit VS2008 or 64 bit Linux both of which are no end
                                        of problems (for ISE).
                                        > Sent via BlackBerry® from Vodafone
                                        >
                                        > -----Original Message-----
                                        > From: Peter Gummer <p-gummer@...>
                                        >
                                        >
                                        > Perhaps the fact that my experience of EiffelStudio 6.2 has been more
                                        > positive than yours, Mark, can be attributed to the fact that I'm
                                        still
                                        > using Windows XP and Visual Studio 2005.
                                        >

                                        Actually I use Win XP sp3 and VS2008 on a 32 bit platform...

                                        In regard of my EiffelStudio GUI problems, the saved window size not
                                        being restored may be related to my screen resolution: 1680 by 1050.
                                        The issue of the "Features" tab failing to update after an F7 compile,
                                        unless I re-open the editor window is easily replicated on my system
                                        but may be unique to my setup. As I said previously these are minor
                                        annoyances rather than major problems.

                                        Out of habit, I tend to operate in User mode rather than as an
                                        Administrator. This opens up a question: should the Metadata consumer
                                        tool work in user mode? Or is there an unknown dependency on running as
                                        an administrator?

                                        Thanks

                                        Kind regards

                                        Mark




                                        [Non-text portions of this message have been removed]
                                      • Chris Saunders
                                        The issue of the Features tab failing to update after an F7 compile, unless I re-open the editor window is easily replicated on my system but may be unique
                                        Message 19 of 19 , Nov 6, 2008
                                        • 0 Attachment
                                          " The issue of the "Features" tab failing to update after an F7 compile,
                                          unless I re-open the editor window is easily replicated on my system
                                          but may be unique to my setup" - I frequently have the same thing happen over several versions now and have made a bug report.



                                          Regards

                                          Chris Saunders



                                          From: eiffel_software@yahoogroups.com [mailto:eiffel_software@yahoogroups.com] On Behalf Of azador1606@...
                                          Sent: November-06-08 6:40 PM
                                          To: eiffel_software@yahoogroups.com
                                          Subject: Re: [eiffel_software] Re: Goodbye to Eiffel?



                                          Any metadata consumer must be able to consume in user mode.
                                          However, that doesn't mean you will have complete access at runtime necessarily - but it would be exceptional not to.
                                          Sent via BlackBerry® from Vodafone

                                          -----Original Message-----
                                          From: "marvin_littlewood_426716" <marvin_littlewood_426716@... <mailto:marvin_littlewood_426716%40yahoo.co.uk> >

                                          Date: Thu, 06 Nov 2008 12:18:21
                                          To: <eiffel_software@yahoogroups.com <mailto:eiffel_software%40yahoogroups.com> >
                                          Subject: [eiffel_software] Re: Goodbye to Eiffel?


                                          Hello!

                                          --- In eiffel_software@yahoogroups.com <mailto:eiffel_software%40yahoogroups.com> , azador1606@... wrote:
                                          >
                                          > Whereas I use 32 bit VS2008 or 64 bit Linux both of which are no end
                                          of problems (for ISE).
                                          > Sent via BlackBerry® from Vodafone
                                          >
                                          > -----Original Message-----
                                          > From: Peter Gummer <p-gummer@...>
                                          >
                                          >
                                          > Perhaps the fact that my experience of EiffelStudio 6.2 has been more
                                          > positive than yours, Mark, can be attributed to the fact that I'm
                                          still
                                          > using Windows XP and Visual Studio 2005.
                                          >

                                          Actually I use Win XP sp3 and VS2008 on a 32 bit platform...

                                          In regard of my EiffelStudio GUI problems, the saved window size not
                                          being restored may be related to my screen resolution: 1680 by 1050.
                                          The issue of the "Features" tab failing to update after an F7 compile,
                                          unless I re-open the editor window is easily replicated on my system
                                          but may be unique to my setup. As I said previously these are minor
                                          annoyances rather than major problems.

                                          Out of habit, I tend to operate in User mode rather than as an
                                          Administrator. This opens up a question: should the Metadata consumer
                                          tool work in user mode? Or is there an unknown dependency on running as
                                          an administrator?

                                          Thanks

                                          Kind regards

                                          Mark



                                          [Non-text portions of this message have been removed]





                                          [Non-text portions of this message have been removed]
                                        Your message has been successfully submitted and would be delivered to recipients shortly.