Read Me File (Epi Info Version 3.3.2, February 9, 2005) Analysis --------- Merging grid views with a field whose name contained an S no longer produces an error message. Two views that relate and have soundex fields of the same name may have an S. Recode expressions of the form x=z can be generated. The type of a defined variable can be changed by assignment so long as its value is missing. MEANS * will produce descriptive statistics (and ANOVA if second variable is supplied), for all numeric variables. The denominator feature of the MAP command will work with the COUNT aggregate. EpiMap ------ Using the Add Points feature of "Vary symbol size for coincident points" and saving as a shapefile, will no longer result in an error box. An invalid map template will no longer cause the program to "hang." All case-based templates will now display correctly. Installation ------------ Uninstalling Epi Info Version 3.3.2 will not remove EpiInfo.INI file from the installation directory. MakeView/Enter -------------- When creating a relationship with a wide table View, the child View may be accessed from MakeView or Enter. Dates with European formats in a grid will be displayed correctly. Read Me File (Epi Info Version 3.3, October 5, 2004) Installation/General -------------------- In previous versions, Administrator rights were required to run the Report Generator, Compact and VisData. Now, once someone with Administrator rights installs Epi Info and shares the installation directory, a profile with User rights can run all the Epi Info applications. In previous versions, the most recently added ODBC driver was not available in the Add/Edit DSN form. This has been fixed. Check code and analysis programs can now use DBCS (mostly Far Eastern) character sets for variable names, program names, file names, string constants, etc. Beginning with this release, Epi Info requires Windows 98 and above to run. Utilization of form caption icons has been made consistent. MakeView -------- In previous versions, non-Western characters were not displaying properly on control captions in a view. This has been fixed. According to the help topic "Handling Date Fields", paragraph 10, date constants in Check code or analysis programs must be in US format. In previous versions, this was not true for Check code. Now it is. In previous versions, date literals in ASSIGN and IF commands were interpreted according to the machine's date setting in MakeView, but treated as US dates in Analysis. Now both programs require US dates. In previous versions, the Create Data Table menu item did not create the data table correctly. This has been fixed. It is now possible to set the default font for the prompts in the View. This applies only to any new questions added, it will not change existing prompts. Please see the help topic "Setting a Default Font." A new property has been added for image fields, Retain image size, which displays images at their original dimensions rather than stretching them to fit the designated space. In previous versions, the user's selection of Access 97 or Access 2000 as the format for newly created databases was not honored. Now it is. It is now possible to use DBCS (mostly Far Eastern) character sets for variable names, prompts, field data, file names, string constants, etc. Note the font chosen must support the character set of the language. In previous versions, dates and numbers could not be added. Now they can. Please see the help topic "Arithmetic (+ - * / ^ MOD)". A new menu item has been added to the Tools menu, Convert Child to Parent. It is active for child views and allows the view to accessed without going through the parent. The related button in the parent will still exist, but when the user clicks on it, a message will be displayed that the association between child and parent has been lost. In previous versions, when the user defined an option box with text on the left, the dimensions and location of the option frame were stored incorrectly. This has been fixed. In previous versions, if a view was open and the user selected File Open but then cancelled, the text "Right Click to Create a Field" appeared on the open view. Now the text does not appear. In previous versions, when an existing table was selected to be used as a code table but the field dialog was not then closed, error blocks could occur. This has been fixed. In previous versions, the field dialog for fields within a grid displayed the Codes button enabled even though it was not functional. Now the button is disabled. In previous versions, if the user created some fields in a view and did not save it, and then attempted to a import a view, the previously created fields were not cleared and appeared in the imported view. Now the current view is saved before performing the import. In previous versions, the user could edit the field name when the legal value or code dialog was open, but that would result in an error block. Now the user cannot edit the field name when the legal value or code dialog is open. In previous versions, the user was not able to save a field in a grid if another field in the grid had the same prompt. Now two field in a grid can have the same prompt so long as they have different names. In previous versions, if the user tried use a existing code or legal table that was a link to different database file, the user received an error block. This has been fixed. In previous versions, importing Epi 6 files with short date (month-day) fields would fail. This has been fixed. In previous versions, importing Epi 6 files with 8-character European dates would be imported with either the 20th or 21st century. Now only the 20th century will be used. In previous versions, if there was little difference in height between two fields. Tab order was not calculated correctly. This has been fixed. In previous versions, the Soundex property was enabled for use in grid fields, even though it was not implemented. In the current version, the Soundex property is disabled for grid fields. In previous versions, if the user created a relate to a new table with the same name as an existing table, sometimes a run-time error occurred. This has been fixed. In previous versions, when Repeat Last was set for a numeric field and the field had data, the format value was not cleared if the data type was changed to Text. This has been fixed. In previous versions, under certain circumstances Insert Page would fail, producing an error block and resulting in duplicate page names. This has been fixed. In previous versions, after deleting the only page of a one-page view, the page name was not set properly. This has been fixed. In previous versions, if the Format Settings menu item was selected and then cancelled after one of the options had been changed, the next time the Format Settings menu item was selected it would display the previously selected options. Now the current settings are showing whenever the Format Settings menu item is selected. In previous versions, performing an Alt-Click on the prompt of a legal value would move the cursor to the top left corner of the view. Now it no longer moves inappropriately. In previous versions, it was possible to type in the Check code editor without picking the variable to which Check code applies. In such cases, the Check code would not be saved even if a variable was selected later. Now the user is not permitted to type in the editor until a variable has been picked. In previous versions, when the user changed a numeric field that had range validation into a text field, a runtime error occurred. This has been fixed. In previous versions, when the user changed the variable type to legal values, the view was not updated and the change was lost. This has been fixed. In previous versions, if the user opened and cancelled the font dialog in the field creation form and then opened and cancelled the default font dialog, a run-time error occurred. This has been fixed. In previous versions, if the user chose the Import Epi 6 menu item but then cancelled, an error block was displayed. This has been fixed. In previous versions, if the Delete Data Table menu item was chosen but then cancelled, the Create Data Table menu item was nevertheless enabled. This has been fixed. In previous versions, when the user moved a option button field, spurious option buttons were created. This has been fixed. In previous versions, the user could type Check code into the program editor without first picking a field to which the Check code applied. This Check code would not be saved. Now, if the user attempts to do so, a message is displayed that that the field should be picked first. In the previous versions, if the user switched pages after adding a new page to a view but before adding controls, the empty page was deleted without warning. Now a message is displayed under these circumstances. In previous versions, when the creation of a conditional relate was cancelled using the X button, the user was no longer permitted to create a relate button with the same name. This has been fixed. In the previous versions the title bar for the Field Definition box was empty after canceling out of a grid. This has been fixed. In previous versions, the message displayed when the user attempts to define a variable that already exists did not get translated. This has been fixed. In previous versions, when the user selected Change Project, the file dialog was displayed with a read-only checkbox. Now the checkbox is not displayed. In previous versions, the Make View From Data Table menu item was enabled even when no view was open, even though an open view was necessary for it to execute. Now this menu item is disabled until a view is opened. Extraneous information has been removed from the About box. In previous versions, the Delete Data Table warning message used the critical error icon. Now it uses the warning icon. Enter ----- In previous versions, non-Western characters were not displaying properly. This has been fixed. In previous versions, when a FIND was based on a date variable, records were not found properly on machines using non-US date formats. This has been fixed. In previous versions, the before and after Check code for the page and record events was not executed when the user navigated to a record using the FIND dialog. In the current version the Check code will be executed when the user navigates using FIND. In previous versions, when the user did a conditional relate based on a comment legal field, the relate button was not enabled even though the condition was met. This has been fixed. In a multi-user environment when the record being edited has been changed by another user, a message with Yes and No buttons is displayed saying that data has been changed by another user. In previous versions, the buttons were not behaving appropriately. This has been fixed. In previous versions of Enter, calling a DLL would cause any subsequent function calls in the same event to fail. This has been fixed. In previous versions, the result of a yes-no-cancel dialog was of text type rather than Boolean (yes-no), with the result that it could not be compared with values. This has been fixed. In previous versions, date fields in grids were displaying wrong century for some date values. This has been fixed. In previous versions, errors in the expression portion of an IF statement in Check code caused both the true and false blocks to be executed. Now, neither will be executed. In previous versions, when Repeat Last was set on a field in a child view, the new field value came from the last record in the table, regardless of the parent record to which it was related. Now the new field value comes from the last record for the same parent. In previous versions, when the user created a relate field with the property Return After One Record set, each time the child view was entered a new record would be created. Now, the same child record will be used each time the user enters the child view. In previous versions, when records were marked as deleted, related grid table records remained undeleted. Now related grid records are marked as deleted. In previous versions, when the user deleted a previously-assigned text field size, the underlying view was not updated. This has been fixed. In previous versions, if there were more than 3 option buttons on the screen, they could not be reached by tabbing. This has been fixed. In previous versions, when Enter was used to compress the database, it sometimes would be lost. This has been fixed. In previous versions, an invalid size error occurred if no size was entered for a text field and the type of control was changed. In the current version this has been fixed. In previous versions, if the code value of a code table began with a space, only the first field based on the lookup would be populated. In the current version, all lookup fields populate correctly. In previous versions, mirror fields could hold only limited text. In the current version, multiline fields will be mirrored into multiline fields. In previous versions, when the user changed screen resolution, the size of option frames would change. In the current version, the frame occupies the same proportion of the screen in all resolutions. In previous versions, if MakeView had been exited by closing a child view without using OK, Enter would display an error block when the Relate button was clicked. This has been fixed. In previous versions, if a permanent variable was set equal to a date field in Check code, the permanent variable was not stored correctly. This has been fixed. In previous versions, closing the select view form with the X in the title bar gives the message "The view is related to a parent view. Please open the parent ..." Now clicking the X button acts like the cancel button. When the user opens a view and changes the page without entering data, a dialog is displayed asking the user whether the record should be saved. In previous versions, if the user clicked on the X button it, an error block would be displayed. This has been fixed. In previous versions, when a form of the dialog command using a database was executed, an error block was displayed. This has been fixed In previous versions, if the user performed a Find and then went to a related view, the search criteria for the Find would be lost. Now the search criteria will be retained until the user does a Find in the related file. In previous versions, when the user deleted a record, any Relate buttons in the view remained active. Now they are made inactive. In previous versions, if there was a conditional relate which depended on the value of the last field on a page, when the user pressed TAB or ENTER control would flow to the next page or record without giving the user the opportunity to work with child records. Now, if changing the last value on a page causes a conditional relate to become active, control flows to the relate button. In previous versions, Enter had a menu item Compress Database. This item is now called Compact Database and its behavior is the same as the Epi Info utility Compact. In previous versions, the Select Table dialog was not being translated. This has been fixed. In previous versions, when the user selected Change Project, the file dialog was displayed with a read-only checkbox. Now the checkbox is not displayed. Extraneous information has been removed from the About box. Analysis -------- In previous versions, dates used in RECODE and GRAPH (Histogram) command generation screens had to use local date format based on Regional Settings, while the command used US date format for syntax checking. Now dates in command generation screens must be in US date format. In previous versions, the user could run other commands while LIST UPDATE was in progress. This is no longer true. In previous versions, when the report generator was run from Analysis, it did not see permanent variables defined or assigned a new value during the current analysis session. This has been fixed. In previous versions, numeric variables could not be used as the outcome variable in logistic regression. Now this is possible. Records with non-zero values will be treated as true, with zero values will be treated as false, and with missing values will be excluded from the analysis. New Analysis commands have been added to delete records, tables and fields and to undelete logically deleted records. Please see the help topics "Delete File/Table", "Delete Records" and "Undelete Records" for details. A new aggregate function, CASE-BASED, has been added to the MAP command to invoke the case-based mapping functionality of EpiMap. For details, please see the help topic "Map". In previous versions, the Soundex code fields associated with text fields with the Soundex property were not accessible from Analysis. Now they are. If the associated text field is changed using LIST UPDATE, the Soundex field will be updated when the window is closed. The Soundex fields of records added using MERGE can be set by running LIST UPDATE. The Analysis REPORT command will now display the report in preview mode if the keyword DISPLAY is used in place of a printer name. In previous versions, when using Epi 6 Direct Read, Epi 6 yes-no variables would be read as string variables with values "1", "0" and " ". They are now read as yes-no variables with values (+), (-), and (.). This means that if they are re-exported to Epi 6, they will be the same as the original file. In previous versions, if the portion of a command before a line continuation character was a valid command, clicking on the first part of the command then on Run This Command would result in only the first part of the command being executed. This has been fixed. In previous versions, when a file which was to be included in Analysis output by means of the TYPEOUT command lacked a tag, the command failed without warning. Now, a message is given. In previous versions, if missing was set on and a variable only had missing values (or only had missing values for one stratum when a strata variable was used), other frequencies, tables and means being run at the same time would fail. Now they will not. In previous versions, multiline variables failed in frequencies, tables and means. Now they can be included, but only the first 255 characters will be used. In previous versions, the Analysis XML output for the report generator did not reflect the setting of the Show Full Prompts option. Now it does. In previous versions, the Analysis XML output for the report generator did not include line numbers as did the HTML output. Now it does. In previous versions, ROUTEOUT caused an error unless a full path was given. Now, the command creates the output file in the directory of the current project if no path is specified or the directory does not exist. In previous versions, attempts to use a DLL method in the expression of an if statement gave the error message "Operators must be of compatible types". Now this no longer occurs. Note that in Analysis DLL methods cannot use expressions involving field variables as arguments. In previous versions, outtables did not function correctly in complex sample routines. This has been fixed. In the past, LIST UPDATE showed "Expr###" as the label for columns containing defined variables. Now it will display the variable name. Note that defined variables cannot be updated using LIST UPDATE. In previous versions, when the FREQ command was used with multiple variables, and the names of those variables were substrings of each other (e.g., FREQ Age Wage), a spurious line with a missing value and 0 count appeared in the output. This has been fixed. In version 3.2 of KMSURVIVAL, specifying a group variable with only one value resulted in unrecoverable error messages. Now, as in earlier versions, a survival graph for all cases is produced. When reading a non-Epi Info data table, a field containing only null, 0, and either +1 or -1 should be interpreted as a yes/no field. In earlier versions, this did not happen if the field contained a null. This has been fixed. In previous versions, MEANS * would sometimes fail. This failure no longer occurs. In previous versions, sometimes logistic regression would not converge to a solution where other products would do so because the initial estimate fell outside the convergence space of the algorithm. Now, the initial estimate is chosen so as to converge to a result where one exists. In previous versions, the Analysis MAP command would not run in silent mode if a template was specified. This has been fixed. In previous versions, the GRAPH command generator screen showed the "Show single cases" button under the wrong circumstances. This has been fixed. In previous versions, it was not possible to read Access and ODBC tables with spaces in their names. Now it is. Due to changes in the underlying database services, some calculations which had previously produced numeric outputs can now produce date outputs, with the consequence that type errors are generated. Now, these calculations have been changed to eliminate the errors while maintaining the previous type behavior. Due to a change in the underlying system functions, strata variables may not be printed. Furthermore, decimal strata variables may not be printed with appropriate precision. This has been fixed. In previous versions, decimal numbers in Epi 6 files were misinterpreted on machines with European number format. This has been fixed. In previous versions, after creating a map, the GRAPH command would sometimes fail. This has been fixed. In previous versions, when no output file was specified and a file already existed with the name built from the information on the Storing Output screen, the output would be appended to the existing file. Now a new output file will be created. In previous versions, the command generation screen for MAP sometimes offered * as a geographic variables even though * is not a permitted value. It no longer does so. In previous versions, the Match command would accept * for the row or column variable when this is not permissible, resulting in an error block. Now these situations result in a custom message and return the user to the Gen screen or program, depending on where the problem arose. The SQL generated for null tests has been changed to reduce the complexity of the expression. In previous versions, the XML report generator output for the complex sample means command contained a standard error which was 100 times that reported in the HTML output. This has been corrected. In previous versions, box-whisker graphs could not be produced for a variable named VARNAME or COUNT. This has been fixed. In previous versions, when running complex sample means, a defined but unassigned outcome variable caused unrecoverable errors. This no longer happens. In previous versions, when using the command generator to generate the GRAPH, FREQ, TABLES, MEANS and advanced statistics commands, it was possible to cause variables to appear more than once in the available variable list. This has been fixed. In previous versions, MEANS * and TABLES * did not work if there were any defined variables. This has been fixed. In previous versions, if the Build Key dialog for RELATE and MERGE was closed using the X in the upper right corner, the Build Key button thereafter produced help. This has been fixed. The handling of errors during the read/import of Epi 6 files has been improved. In previous versions, attempts to use the MERGE command when the READ table was not an Epi Info or MS Access type caused a series of error messages. Now, attempts to merge data into unsupported file types results in an explicit error message to that effect. In version 3.2.2, the SUMMARIZE command did not work with the COUNT() aggregate. This has been fixed. In previous versions, it was not possible to omit the path when reading an Epi 6 rec file in the current project directory. This has been corrected In previous versions, errors occurred with some file types when writing a defined variable set equal to UniqueKey. This has been corrected. In previous versions, Excel files with a name part longer than 8 characters could not be read. Now they can be. In previous versions, after reading a non-Epi Info table, such as an Excel worksheet, the command generator for RELATE and MERGE displayed other data sources of the same type instead of the current project tables. This has been fixed. In previous versions, reading a table not part of an Epi Info View could cause an overflow error. This has been corrected. In previous versions, in Complex Sample routines, if the PSU variable was a defined variable set to missing, an unrecoverable error would occur. Now this produces a single error message. In previous versions, specification of a defined but unassigned variable for SORT would not produce an error message, and subsequent LIST and WRITE commands would fail. Now, SORT will produce an error message and subsequent commands will not fail. In previous versions, after generating a GRAPH command for a hi-low graph, the command generation screen for graph was not displayed properly. This has been fixed. In previous versions, running MEANS * with a table that contained no numeric variables would produce an error block. Now the same circumstances will produce a customized error message. In previous versions, reading an Epi 6 table that included a field named uniquekey and writing that table in Epi6 produced an Epi 6 table that could not be read correctly. This has been fixed. In previous versions, reading an Excel 3.0 or 4.x worksheet and then writing it in the same directory in the same format sometimes failed with an error message that the database was read only. Now this has been fixed. In previous releases, it was possible to specify a defined but unassigned variable as the PSU variable for Complex Samples without receiving a specific error message to that effect. Now a message to that effect is displayed. In previous versions, it was possible to specify text variables in the MATCH command without receiving a specific error message to that effect. Now a message to that effect is displayed. In previous versions, when the user clicked on the New button in the program editor, all the commands in the editor were cleared. Now a message is displayed to verify whether the user wants the commands cleared. In previous versions, when generating a READ command using delimited files on a machine where Excel was associated with those file types, a hidden preview window would be opened, causing the read to fail because the file was in use. Now, the file preview window for these file types will not use Excel and will not be hidden. In previous versions, some error messages relating to the DIALOG command did not match the vocabulary of the command generation screen. This has been changed. In previous versions, the LIST UPDATE command did not change the cursor to indicate busy status. This has been changed. In previous versions, a change was made to add single quotes around table names in the READ command generator when the data source was ODBC. This has now been extended to RELATE and MERGE. In previous versions, there were situations where output was not being translated. This has been fixed In previous versions, Epi 6 direct read of file whose extension is not REC would create file of the same name with a rec extension which is not an Epi 6 rec file. In previous versions, the print button on the output window remained depressed and disabled for 30 seconds after clicking to enable the printing to take place. Now the delay is only 20 seconds, during which the image and caption change to indicate more clearly what is going on. In previous versions, when generating a READ command, if there were no Epi Info views in a project, an empty list would be displayed. Now the list of all tables will be shown. EpiGraph -------- In previous versions, the moving average was not displayed properly on Moving Average type graphs when there were missing values in the data. This has been fixed. In previous versions, graphs were not plotted for missing values of the strata variable even when Include Missing was selected. Now they are. In previous versions, there was a Display Data menu item. Now this item has been removed. A more complete version of this feature can be found in the Customization dialog. In previous versions, date variables were displayed in numeric format in scatter XY and step graphs. This has been fixed. In previous versions, when a pyramid graph had a cross-tabulation variable with only a single value, a message was displayed that a cross-tabulation variable was required, but it would precede to generate an appropriate graph. Now the message has been removed. In previous versions, when a graph template was created with fewer x-axis points than the number present in the data being graphed, the additional points would be displayed in an area which could be reached by use of a horizontal scrollbar. Now the scrollbar has been removed and all points are displayed in the visible portion of the graph. In previous versions, the Y Axis would sometimes become crowded with labels and tick marks. This problem has be fixed, except for Horizontal Bar and Pyramid type graphs. In previous versions, the Hi Low graph would in some cases result in a series of errors. This has been corrected. In previous versions, graph background colors were not stored in the graph template. This has been fixed. In previous versions, when a graph template was created from a graph with a strata variable, strata information was being saved in the subtitle of the graph template. Now this is not done. In previous versions, when a graph template was created from a graph with a strata variable, the save template dialog was displayed once for each page. Now it will display only once. In previous versions, Y Axis labels on Pyramid graphs were partially blocked out. This problem has been fixed. In previous versions, if a template was used that had been created with a smaller number of points on the horizontal axis than the current data source, the additional points could only be seen by use of a horizontal scrollbar that did not appear in the Analysis output window. Now all points in the current data source are displayed without a scrollbar. In previous versions, the Set Y Axis Style menu item was available for Pie, Polar, Scatter 3D and Pyramid graph types, even though the setting is not applicable to them. Now this menu item is disabled for these graph types. Also, the Set Y Axis Style will not be accessible when the Y Axis labels are dates. EpiMap ------ In previous versions, a dot density rendering could not be displayed over a case-based rendering, nor could such a map be created through a template. This has been fixed. In previous versions, when data was added to a layer and it was displayed as something other than a choropleth, the saved template would still have a choropleth setting. This has been fixed. In previous versions, when multiple layers used the same shape file the map would not be displayed properly. Now the order of the layers in the map manager determines the display order, while the currently selected layer determines which legend is drawn. In previous versions, if Advanced Labels were added to a map and set to a large size and the map saved as a template, the label size would not be restored properly. This has been fixed. In previous versions, when a Dot Density rendering was displayed over a Unique Values rendering, the Unique Values display and/or its legend were not saved in a template file. This has been fixed. In previous versions, when multiple shape files were used on different layers and their data was represented in multiple ways, a variety of errors occurred. This no longer occurs, although there still may be particular combinations which fail. In previous versions, when Add Points was cancelled after selecting a data source for the point data, the data source was not removed from the map. This has been fixed. In previous versions, if a map created from Analysis specified both a title and a template containing a title, EpiMap would use the template's title. Now, any title from Analysis will override the template's title. In previous versions, the Layer Properties dialog tabs were not translated. This has been fixed. In previous versions, after creating a map from Analysis using a template and resaving to a new template with changed individual colors, the loaded choropleth map turned black. This has been fixed. In previous versions, when producing a map from Analysis, if the geographic fields were of different types, an error block would be produced. Now a customized data type mismatch error message will be displayed. In previous versions, if two Choropleth renderings were done on two shapes files and the ramp colors or number of choropleth classes was changed for one of them, the changes would not be saved to a template. This has been fixed. In previous versions, if a dataset failed to join with a shape file when generating a Case-Based map, the Case-Based button would be disabled even though no map had been displayed. Now, the Case-Based button remains enabled if generating a Case-Based map fails. In previous versions, if the shapefile had no text field or the data table had either no text field or no numeric field, a message was displayed and thereafter it was impossible to select another table. This has been fixed. In previous versions, if the user created a Unique Values rendering when a value for a Unique Layer was missing, a run-time error would occur. This has been fixed. In previous versions, if a language other than English was used when creating a shape file, standard fields such as "name" and "area" were saved with the wrong names. This has been fixed. In previous versions, map templates were not saved properly on machines using non-US number format. This has been fixed. In previous versions, clicking on other property tabs immediately after applying a dot density with overlay would cause a runtime error. Now the runtime error will not occur, but there are still situations where an overlay on top of dot density will not save properly. In previous versions, selecting the Background Image menu option when there was no background image or when loading was cancelled or failed caused a run-time error. Now this menu item is disabled in those circumstances. In previous versions, when points, lines or shapes were added to a map, or when case-based mapping were used, typing an invalid symbol name or size would result in an unrecoverable error. Now, in these circumstances, valid default symbols and sizes are used. In previous versions, the Add Points button would remain enabled after the user added data to the layer. Now it is disabled. Epi Report --------- In previous versions, when creating a new DSN in the Report Generator, help was sometimes not available. This has been fixed. In previous versions, if the user printed or previewed the report, the row data was sometimes getting cut when the table spanned multiple pages. This problem has been fixed for most of the tables. For the tables that fail, an option is still available to print using the functionality provided by version 3.2.2. Besides this, there is a new feature to insert a page break at the desired position on the template. In previous versions, when two or more elements overlapped in the template designer, they became disconnected when the report was generated. This has been fixed. In previous versions, the tree did not display the newly added nodes (line listing and pivot table queries, analysis table and field aggregates) immediately after the user added them. Now the tree expands and highlights the new nodes in bold. In previous versions, there was no way to know whether any new commands were added to an existing Analysis XML file. Now the designer displays the command selector dialog when new nodes are found. In previous versions, fonts not based on the Latin character set were not fully supported. Now they are. In previous versions, there was no way the user could terminate the report generation process. Now a cancel button is provided, which will return the user to the designer. In previous versions, after a table pair was selected in the relate tab of the query builder, there was no way to tell the table to which a field list belonged. Now each field list is labeled with the table name. In previous versions, there was no way the user could hide or move the toolbar buttons. Now they are categorized by functionality and split into different groups which can be hidden or placed separately. Also, another group has been added for setting the width or height of the selected element without invoking the property window. In previous versions, the only way to add vertical space to the bottom of the template was through CTRL+J. Now a button to do so has been added in the toolbar. In previous versions, special characters and symbols were not allowed in query names. Now they are. In previous versions, when the print menu item was chosen after generating a report, the print dialog was hidden. This has been fixed. In previous versions, when the print toolbar button was chosen after generating a report, the print dialog was displayed. Now it is not. In previous versions, an unnecessary message was displayed when editing a line listing or pivot table query consisting of memo fields. Now it is not. In previous versions, when adding a table from Analysis, formatting changes such as border styles and colors were not being retained. Now they are. In previous versions, line listing output at the grand total level showed the name of the aggregate plus the word "$Grand." Now only the aggregate name is shown. In previous versions, clicking the Clear button in the query builder screen for pivot tables caused an error message. This has been fixed. In previous versions, if data for line listing and pivot table output contained valid HTML, it was interpreted and rendered as such in the table cells. Now the data is treated as plain text. Menu ---- In previous versions, when a variable was defined but not assigned and then was used in macro substitution (@@), the program would hang. This has been fixed. Data Compare ------------ In previous versions, opening an invalid script file would stop the current data comparison. This problem is fixed. In previous versions, the field comparison number and percent values in the comparison report were incorrect. This has been fixed. In previous versions, the Data Compare statistics report showed the percent difference only to the nearest percent. Now it will display as many as 4 decimal places. Epi Lock -------- In previous versions, if the user entered a wrong password when decrypting a file using 40-bit encryption, no message was displayed. Now a message is displayed. In previous versions, passing an erroneous command line parameter would result in duplicate error messages. This has been fixed. In previous versions, passing an erroneous command line parameter would leave a hidden copy of the program running. This has been fixed. Tbl2View -------- In version 3.2, Tbl2View produced views which could not be updated with Check code. Now, when opening views in Enter or projects in Analysis, the structure of the views will be updated to the current version. In previous versions, Epi Info fields were added to the table even if the operation was cancelled. Now this does not occur. In previous versions, code tables were not converted correctly. This has been fixed. When the user chooses a table name containing a space, the program prompts for selection of a different name. In previous versions, if the replacement name matched the name of an existing table, a runtime error occurred. Now a message to that effect is displayed and the user is prompted to enter a different name. In previous versions, choosing an Jet reserved word as a field or table name caused a runtime error. Now the user will be asked to change the name when a reserved word is discovered. NutStat ------- In previous versions, reports could not be generated when the data table had a space in its name. This has been fixed. In previous versions, during a batch update, BMI, MUAC and Head Circumference statistics were not added even when the reference was set to 2000 growth curves. This has been fixed. In previous versions, the options screen had a label reading "Weight For Height". This has been corrected to "Height For Weight". In previous versions, a runtime error occurred when a report template was run against a different data table than that in use when it was created. Now a report template can be used regardless of what table was used when the template was created. In previous versions, inappropriate help was displayed when the user pressed the F1 key on the main screen. This has been fixed. Help ---- The help section for the HELP command was improved. ReadMe (Epi Info, Version 3.2.2 April 14, 2004) General ------- Newly created .MDBs will now be created according to the database version set in Menu. The window displayed when executing the Dialog command will now adjust its size according to the length of text for the prompt. Analysis -------- The feature which permits the user to specify a database on the Analysis command line and have it used as the default now functions correctly. In prior versions, if the directory path to the current project included any spaces, maps would be generated but not displayed in the Analysis output. Analysis/EpiGraph ----------------- In prior versions, when the GRAPH option TITLETEXT was used, the title string appeared both as the overall title over the graph image(s) and as the main title of each graph. Now, if multiple graphs are being created (i.e., a STRATAVAR is specified), the TITLETEXT string will be displayed only as the overall title. If the TITLETEXT contains the pipe character ("|"), the string to the left of the pipe will become the overall title and the string to the right of the pipe will become the main title of each graph. If only one graph is being created (i.e., there is no STRATAVAR), the TITLETEXT string will be displayed as the main title of the graph. If the TITLETEXT contains the pipe character ("|"), the string to the left of the pipe will become the main title and the string to the right of the pipe will become the subtitle. These titles will override any specified in a template, if one is specified. The legend will now display even if there is no cross-tabulation variable. In prior versions, null stratification values were not graphed. However, a limitation remains still for Scatter XY, Hi-Low, Box-Whisker, and Step graphs. When graphing with a weight variable, error messages were produced if the weight variable had missing values. Now records with null weight variable values are skipped. A new date format applicable to time-only values has been included in those provided in the GRAPH command generator screen. It should be noted that users are not restricted to the provided formats, but can enter their own custom format. This capability does not include Scatter XY, Hi-Low, Box-Whisker, Step and Scatter 3D graphs. An error message stating that a cross-tabulation variable is required for a Stacked Histogram graph has been removed. Graph will now display a Cross-tabulation (CROSSTAB) and/or strata variable (STRATAVAR) legend when there is only one value. Graph page title is now centered. Graph images will now consistently be sent to the output window. The EpiGraph menu has been slightly changed for improved usability. Installation ------------ OfficeSetup is now executed from Setup to fix a known Microsoft Office problem with executing DHTML edit control (used by Epi Report). See This had been removed from Version 3.2 as it was thought to be no longer necessary. When Epi Info, Version 3.2 was uninstalled, the "Open Link in New Window" option (when right-clicking on a link) did not work. When Epi Info, Version 3.2 was uninstalled, the Add/Remove Programs from Control Panel states that there are no programs installed on the machine. Upon scrolling down, the programs would appear, but the buttons were all grayed out. Epi Info now installs MDAC 2.7 Refresh to handle a known issue with Microsoft SQL Server service pack 3. A detailed description of the issue is available at . Epi Graph now runs properly the first time the Graph command is executed from Analysis. An Epi Report error message is no longer displayed as soon as OK is clicked in the MDB selection dialog. On a machine with MDAC 2.7 SP1 uninstalling and reinstalling Epi Info, Version 3.2, causes all the applications that use ADO (Advanced Statistics in Analysis, Epi Map and Epi Report) to throw errors. This does not happen with version 3.2.2. MakeView/Enter -------------- DIALOG commands which assign values to a variable will no longer produce an error message. Check code will now run in multiline variables. An error message will no longer display if clicking on the Program button and then clicking on Insert Page. Table-to-View ------------- In prior versions, if there was a space in the table name, Table-to-View would automatically replace them with underscores in the view name but not the table name, causing problems with the table. Now, the user has the opportunity to rename the data table if it does not have a valid Epi Info name. In prior versions, if the user selected an empty database, the program proceeded to the next screen. Now, the user cannot continue if there is no table available. In prior versions, the View table was created with the Check code field as Text instead of Memo which limited the amount of Check code which could be entered. In prior versions, View tables were permitted to be converted to a new View. Table-to-View will no longer show View tables in the table list. A field called ID will no longer be created when converting from a table to a View. Installation Procedures ----------------------- How to Free Up Disk Space on Your Hard Drive -------------------------------------------- The Setup for Epi Info is now using the Windows Installer. As in the case with all advances in technology, Epi Info Setup requires more space and time for execution. Even if you do not intend to install Epi Info on drive C, the installation may require about 260 megabytes (MB) of (mostly) temporary space; about 200 MB is freed up after the installation. If Setup indicates that there is not enough space on the disk, try the following: 1. Close ALL applications. 2. Search/Find C:\*.tmp and delete all *.tmp files found on drive C. 3. Empty the Recycle Bin. 4. Compact any large MDB files. 5. Move large files to another drive (D drive or LAN drive). 6. Zip large files remaining on drive C especially word files or .mdb's. 7. Go to Start | Settings | Control Panel | Add/Remove Programs: a. Remove all unnecessary programs - especially previous versions of Epi Info; and b. If the machine has a drive D hard disk, “Remove” some programs and re-install those programs on drive D (making certain that the appropriate installation medium is available). NOTE: If the above steps do not create enough free space, and the target machine has a D drive, try the following: To set the temporary directory in Windows 9x perform the following steps: 1. Edit the autoexec.bat and add the following lines: Set TMPDIR=D:\temp Set TEMPDIR=D:\temp 2. Reboot your computer. To set the temporary directory in Windows 2000/XP perform the following steps: 1. Right click My Computer. 2. Click Properties. 3. Click the Advanced tab. 4. Click Environment Variables. 5. Highlight the TEMP variable. Click edit. Enter D:\TEMP. 6. Highlight the TMP variable. Click edit. Enter D:\TMP. 7. Reboot your computer. Repeat steps 1 - 4 and 6 above, with suitable changes from C to D. ReadMe (Epi Info, Version 3.2, February 4, 2004) Analysis -------- Date values in the INI file (permanent variables) are now recognized. Menu -------- In prior versions, it was not possible to pass in multiple command line parameters when using the EXECUTE command. ReadMe (Epi Info Version 3.2 January 22, 2004) General ------- On machines that have a non-US number format setting, numbers can be entered into dialog boxes produced by the DIALOG command using either the US decimal point (a period) or the machine's local decimal point setting found in Control Panel/Regional Settings (normally a comma). Date values in the INI file are now being stored in local format, so when INI files are exchanged between machines that use different date formats, the INI file must be manually modified to express dates in local format. On machines that have a non-US date format setting, dates must be entered into dialog boxes produced by the DIALOG command using the machine's local date format as found in Control Panel/Regional Settings. Analysis -------- In prior versions, it was possible to specify a defined but unassigned variable as the PSU variable for Complex Samples without receiving a specific error message to that effect. In prior versions, it was possible to specify text variables in the MATCH command without receiving a specific error message to the effect. In prior versions, while using the TYPEOUT command when a file lacked the tag, the command failed without warning. A message is now displayed when a file lacks the tag using the TYPEOUT command. When using Epi 6 Direct Read in prior versions, Epi 6 yes/no variables were read as string variables with values "1" and "0". They are now read as yes/no variables with values (+), (-), and (.). This means that if they are re-exporting to Epi 6, they will be the same as the original file. In prior versions, the number of degrees of freedom reported for matched (conditional) logistic regression was one too high, which resulted in probabilities higher than that which should have been shown. In prior versions, when FREQ, TABLES or MEANS was run with a defined variable as the weight variable, a spurious error message, "Weight must be numeric variable", appeared. In prior versions, the logistic regression procedure incorrectly determined that it was failing to converge and stopped computation with a message to that effect. Menu ---- An OnBrowserExit command has been added. The OnBrowserExit command is executed when the user clicks the Close button on the browser window that is generated from Menu. This command works similar to the StartUp and ShutDown command. The string IniPath and InstallDir in the Epi Info Registry now includes a backslash. This change will not change the Epi Info system variables, IniDir and InstallDir. Epi Report ---------- The Line Listing Group feature now allows for the spanning of cells of the group row. The header “EpiRepGen” that was displayed in the Windows Task Manger dialog box has been changed to display “Epi Report”. The prefix and suffix of date values in the SELECT tab of the Query Builder has been changed from single quotes to the ‘#’ sign. The Footer row in Line Listing now spans all the cells after new columns are added to the existing query. Borders now appear if the fetched value is null for any of the cells in Line Listing. The correct value is now fetched for the grand total that is defined at the bottom of the report. The Office-CD Fix menu item has been removed from the Epi Report Help drop down menu. The new installation routine has eliminated the need for this feature. The performance speed of Epi Report has been improved in several areas including report generation.