[EpiData-list] New (experimental) version of rewritten Analysis 0.8.4.0

EpiData development and support epidata-list at lists.umanitoba.ca
Fri Oct 13 16:05:55 CDT 2017


Today a public test version of rewritten Analysis was made available for 
Windows and Linux. (v0.8.4.0)

Test versions are available as packed archives from: 
http://epidata.dk/testing.php (at the bottom)
Extract the files provided and maintain the folder structure.

We are rapidly approaching a version, which can be used in the 
management for real projects.
But for a while still please only use for testing.

Start epidataanalysis (exe) and see help menu "Commands Overview". The 
file shown is commands.html . Also study the 
"How_to_adapt_your_pgm_file_to_new_EpiData_Analysis_format.pdf" found in 
the docs folder.

Our aim is to change as little as possible in the final syntax, but some 
changes are necessary due to demands when we write the software for more 
operating systems (currently Linux and Windows - Mac version will come).

The main changes to the command syntax are:
   1. use ; at the end of a command in a pgm file
   2. use := to give variables a new content      . e.g. v1 := 10;
   3. options are indicated with ! instead of /,  e.g. freq v1 !cum ; 
means age !t;
   4. filenames must contain extension, and enclosed in "projectmain.epx".
   5. Similar aspects are grouped, e.g. list data; list variables; list 
datasets
   6. Defining and editing variables, labels etc have a common 
structure: new var; edit var; drop var;
   7. Many additions allow comprehensive programming

A major strength of the new analysis is that you may read and save 
datafiles in csv, dta (stata), epz and epx format directly. You may also 
read the classic rec files. All executed commands are saved in a pgm 
file defined by the user. Reading, editing and saving now supports 
encrypted data, user logged data and files containing related datasets .

We know of some issues and bugs in the software released today, but 
decided to release at this point, since we need more input to remove all 
bugs at the detailed level. We have almost 300 validation tests and only 
one special error. Comparison with standard estimates show precision 
with difference from correct result at -2.06197786709744*E-12 . This is 
a very promising precision. Examples of known issues are that cycle is 
not corrected after modifying data, or errors in special types of "for 
....to..." loops.

If problems arise please report to this list. But make sure you WORK on 
COPIES OF DATA.

Commands  and statements added recently :
* "append" and "merge" handles related datasets
* "check" command with 4 sub commands: data, key, relate or study
       (document data content, e.g. "check key id visitdate" will tell 
if there is a unique key for id+visitdate)
* "reorder" variables command
* "reset" command
* read and save files with extended login and user protection defined.
* drop data , including remove records marked for deletion
* control of fonts, including ability to set your own default by saving 
to "startup.pgm"
* "set" options added recently:
  - EDITOR and OUTPUT FONT + SIZE: to adjust fonts and size
  - OUTPUT VIEWER: controls what type of output is generated. text or 
html is supported
  - INCLUDE DELETED: controls if records marked for deletion is part of 
the dataset
  - SET FORMAT VARIABLE LABEL and
  - SET FORMAT VALUELABEL: controls default option for value / variable 
text

Many other issues managed:
* enhanced and informative feedback in more situations (more indirect help)
* run commands in "startup.pgm" on initiation of the program. (Placed in 
same folder)
* Dragging files (data or .pgm) onto the program either opens the file 
or executes the .pgm
* Variables can now be expanded using modifiers like "*", "?" or "-" , 
e.g. list variables v*
* Extended editor with more functionality
* F1....F12 function keys work as in Classic EpiData Analysis (except F2)

Please test and challenge function.

Jens Lauritsen and Torsten Christiansen
EpiData Association
Denmark



More information about the EpiData-list mailing list