way to keep a trace of the delopment of a model

[LAUJJL]

Hi

I have a methodological problem.

When I develop a model, I start with a highly simplified reality, make a model of it and work with that model, doing everything that can be done with it as if it was the last version (analysing, finding different policies, sensibility analysis, optimization, games etc.). Once this has been done whatever the time it takes (days, weeks or months), I enrich this simplified reality to make it a little bit closer to the reality itself. And I start the same process again etc..

I fact at each step, I am working as if the reality was as simple as that and that I had to find policies that can be implemented to manage this kind of reality.

I follow strictly Coyle's advice: 'the key to successful modeling is to keep one's understanding of the model and what it says about the problem ahead of its size'. The only difference with my way of working is that I change 'ahead of its size' by 'way or far ahead of its size'.

Working this way I want to keep a trace of it and it is not sufficient to keep a version of the model at each step of its development. I want to keep a trace of the main works done with the model, so that I can easily go back to that step and review what I have been doing at this stage, even if it is months later.

I have looked at the Venapp which looks very interesting as it is adapted to Vensim.
The question is that I am sometimes obliged to build batch models from VBA Excel using the dll functionnality, to automatize long and tricky batch works.

Is it possible from an application written in Venapp to call an application written in VBA excel using the dll functionality?

Or to use the dll functionnality, I am obliged to build all the interfaces in Excel?

I have too looked at the Sable which seems to have a procedural language that permits building batch works more easily. Is it possible to launch VBA excel from Sable or is it possible to build the batch works from Sable without the need to work with the VBA dll?

I have decided to be much more strict about the way I work, because it is generally impossible for me to review past works correctly even if I have kept the models used.

Any suggestion is welcome.

About the way I work, I do not think that many people work this way. I can do it because I am at the same time, the modeller, the decision maker, the problem owner and the man who will have to implement the policies found.

Regards.

JJ

[tomfid]

One option would be to adopt a source control system like Subversion or Git. Each time you check in a revision, you can enter a comment, which makes it fairly easy to document your work.

[LAUJJL]

Hi Tom

Thank you for your answer, but I am not sure you that you understand my problem.

I do not only want comments on why I made some modifications of my model.

If I make a revision to a model, I can just change the name of the version and comment the model
in the model settings. I can too comment a run in Vensim.

What I want to keep is a trace of how I have been working with the model to get new ideas, policies, new insights, using the different tools available. In the course of development, it is hard to spot what I have been doing even some days before, even if I have different version of models with their comments. I am looking for something that could be used even by somebody having no knowledge of modelling. If I go back to something I did before, I prefer to concentrate on the problem itself than on a model and its technicalities, like anybody concerned with the problem and having no knowledge of SD precisely.

I repeat that in the course of the development of a model, I spend a lot of time, working with each version, not only to test it, but to take from it every thing that can be done. This helps to chose the way that the model will be built later on, but the deep understanding of how the model works.

Best regards.

JJ

[tomfid]

In that case, I don't have a ready substitute for the old standby, a lab notebook.

My own habit is to keep the model and its associated files in a separate folder, with a version number in the filename for both. For minor revisions, I save the model with a new version #. From time to time, I clean up the files in the folder and make a fresh copy with a new #. I dedicate one view of the model to a todo list and comments, and I maintain a parallel Word document for logging observations.

For some projects, we use svn, because it's convenient for rolling back changes and syncing files among our team.

[LAUJJL]

Hi Tom

Thank you for your answer.

Regards.

JJ