During recent months extensive work was done rewriting core parts of the software to a unified form for Entry and Analysis regarding reading, saving, exporting, importing and validating data in the internal structures. All of this is not visible to the end-user, but remedies some of the problems seen with different functioning of Entry and Analysis. The strategy is to ensure unified modularity of the internal structure instead of the "add lumps" principles which to some extend have been used over the first five years of development.
At the same time known problems and deficiencies are fixed, e.g.: a. In Analysis the graph and table dialogs cannot be resized b. In Entry a clearer logic of creating data files and chks can be implemented, and implementation of Good Clinical Practice principles.
As part of the core parts development a key highly technical decision has to be taken: One subtle aspect of earlier standards has been that integer variables were treated in a different form depending on the 1-4 or 5 and larger. Where integers of length 5 (#####) and larger actually were kept internally in the software (Entry and Analysis) as float variables with 0 decimals. The reason for this rests in the old Epi6 format and memory usage in the 1980'ies. We have taken the decision that from next versions of Analysis and Entry all will be built upon a common core system, where all integers (up to 14 in length) are treated as integers internally in the system. The change will only have impact for users reading rec files in other software than EpiData. All existing rec files will be read with no changes. For exports and imports the software will adapt appropriately. But users will have to use the most new versions for reading files created after the change.
If anyone reject to this decision please discuss the consequences here on the list.