Change Log

Known issues as of version 1.1.1:

Sometimes if you clear a segment and load a segment it will show a name conflict. Save the form and open it up again to get around this.
Copying an image using the "copy" or "cut" command in FormReturn between two different segments or forms will cause the new form or segment not to load properly. Work-around: Don't copy or paste images between forms or segments, load them instead.

11 July, 2010 - Version 1.1.1

Enhancements:

Improved mark aggregation rule. Will now have a rule for no checkboxes selected and one for any number of checkboxes selected over a certain number.
Improved reprocessor screen, reduced steps required.
Modified the lowest value for the mark threshold to enable negative thresholds. From -10,000 to positive 10,000.

Bug Fixes:

Fixed reprocessor screen where it was not clearing the error flag.
Fixed export of existing publications where sometimes an embedded image would cause the export to fail.
Fixed export preferences where the include processed times flag was being ignored.
More PDF files will detect properly.

21 June, 2010 - Version 1.1

New Features:

A very big change to the checkbox detection algorithm. Single checkboxes and complete checkbox selection will now work. The new detection algorithm as a result is more robust, where you can also now have rectangles or any other artefacts around the checkboxes without it afecting the detection process.
Reasons are now given for forms not detecting properly, viewable in the unidentifed images panel. For instance, if a Form ID barcode is not found, or a segment barcode is missing.
A form processing progress indicator has been added to FormReturn server. This will allow you to see when a form gets processed and how long it takes to process a form. It will also attempt to show how much time is remaining before all forms are processed.
A form is no longer invalidated if one or more checkboxes are invalidated - it will now show an error count.
Users can now manually select the program language in the general preferences and in the server's main screen.
Can now specify the read direction the OMR reader reads combined character OMR grids (to produce names, IDs etc). Now ofers the choice of left to right or top to bottom.
Added a new "Assistance" button to the top right of the program's main window which will allow users to have a reseller contact them for support or sales. Will also appear when the program issues a trial license and when the trial license runs out.
Snap to grid is implemented for snapping editor elements to a grid. You can enable/disable snap to grid as well as adjust the gird size and if the grid displays or not. Assists with form design and alignment.
All program errors are now recorded to application.trace and server.trace in the FormReturn data directory.
FormReturn will now accept PDF fles as an image type for processing captured form images.
A "Reprocessor" screen has been added to both the unidentifed images panel and the captured data -> form page panel. This feature will allow you to manually process forms (particularly in the case of a form not detecting properly), or modify the data that was captured (in the case where the form contains errors or did not correctly capture the data).

Bug Fixes:

Database connection pooling was added as it wasn't relised that OpenJPA creates a new connection for each query it performs. This became an issue for people who tried to publish or export past around 700 forms because their operating system would limit the number of open ports and therefore cause an issue with connecting to the database. DBCP has been added to prevent this issue from ever happening again.
Fixed a large issue with OpenJPA/Apache Derby where a CLOB/BLOB bug would often appear if a blob doesn't have a lazy load annotation. Suspect this was always the cause of any CLOB/BLOB error messages people had in the past.
A bug existed where if the output folder in the folder montor contained an image with the same fle name as the one being processed in the input folder, the output folder fle would not be overwritten and cause the input folder fle to continuously process.
Some barcodes, particularly ones that start with 0 were not correctly renedered (due to the barcode creation library), causing forms that have more than 4 segments not to detect properly.
Some checkboxes and text areas were having their text clipped (or not appear at all). If a font was larger than the allocated space it would not display. This has been fxed by disabling clipping in the rendering of forms.
Some asian font fle names were not reading correctly, this was an error with the font reader not providing the correctly encoded data.
If a font exists in more than one place, it caused the fonts to double register, leading to many font problems. A font will now only ever be loaded once.
When testing the second or third page of a form that had already been published in the test preview area of the form editor, the segments would not be detected due to a mismatch of segment barcodes.
A bug existed where publishing a form template on a set of data that already had a publication on it would cause issues.
The software update dialog was incorrectly saying there were program updates in some instances. We suspect this may have been caused by proxy servers providing messages and they were being confused with version information.
The retain shape feature in the image element in the editor caused a bug if no image was selected and retain shape was turned on.
A barcode would resize past the borders of the page if dragged from the bottom left corner. This would lead to a number of issues.
An internal issue was fxed in the code where a JPA entity manager was not being closed/cleared leading to various data matching issues.
Re-exporting a template publication was not working properly, would export like a Form ID publication rather than just the template.
Fixed a cast error on the source data table's feld name column order editor.
Changed the way data is captured, previously stored captured data as a serialized multidimensional java string array - this was not good for many reasons. Have added the CHECK_BOX table to the database and re- created a similar scenario in an SQL friendly method.
The status message bar at the bottom right of the programmer had an issue with the timer that would compound active threads when messages were shown (would keep looping the timer thread rather than quitting once frst run). This caused a large amount of unnecessary thread activity.
Removed some static felds that were causing memory leaks.
Fixed an issue where publishing over 800 or so forms would cause the program to run out of memory. Forms were frst rendered to memory and then outputted as PDF fles. The system will now render a form and output it as a single pdf fle, then merge all of the PDF fles together (in the case of form colation). Up to 3200 form page publications have been tested without issue.

Enhancements

When switching between panels, if you go back to the source data or captured data panels it will return to where it was before you switched.
When opening some dialogs, the default focus is now correctly selected on the text areas and any associated areas.
When switching between tabs, the program will refresh any of the data in the list so it shows the most up-to-date data. Particularly useful in the cases where you may add a publication, but already have the publication tab open and a refresh used to be required to show the new publication.
Forms are pre-rendered before they are displayed to the screen. This means that the loading dialog will not disappear until the form has completely loaded.
Fixed the control-c (command-c on mac) issue where if you did a copy or paste with a text area, it would also add c or v to the text area.
If an image is exessively large, the program will automatically downsize the image as excessively large images use excess storage space and take longer to process. You will be notifed of excessively large images in the new server processing statistics in the server.
Fonts will now be listed for text areas in your own language if the font contains a localized translation of the font name (if your language is in the language list in general preferences and it is selected). This is particularly useful for those people who don't speak English and aren't used to English font names.
The source data record editor is very much improved - no longer requires updates and works similar to the checkbox editor.
The form publisher has been enhanced with less chance of a publishing status dialog blocking the application if a publicaiton fails.
Added a form page processed date column to the data export. Date is exported as Year, Month, Day, Hour, Minute, Second for easy column sorting in spreadsheets.
Fixed some minor dialog size issues for Windows 7.
The name of the fle that was processed is now stored in the PROCESSED_IMAGES table.
The publication type is now displayed in the captured data -> publications panel as well as the publish screen.
Error messages and notifcations are now shown in a scrollable dialog box so they don't extend past the width of your screen.
Published form pages are now ordered according to the order of the source data records. This was random in the past.

January 28, 2010 - Version 1.0.7

New Features:

SPANISH LANGUAGE INTERFACE SUPPORT - Will automatically switch to Spanish for anyone who has that language selected for their operating system.
Internationalization Support - You can translate the interface into any language by modifying the language.jar file.
Ability to edit field names, mark order indexes and aggregation rules AFTER a form has been published. This feature is part of the publication settings dialog in the captured data manager.
Ability to export and import an answer key file (CSV format). Works in combination with the publication settings.
Barcode Detection through the new Barcode Reader tool - Will detect most major 1D barcodes as well as QR Code.
Barcode ID reconciliation - Will reconcile barcodes with a source data record, much the same way that OMR ID Grids reconcile. Particularly useful for healthcare industry where barcodes are often affixed to forms after they are printed out.
Ability to change the default font of the program using your own custom TrueType font file. This will also change the font of the checkboxes and the barcode font.
Clear all data feature added for publications and forms in the captured data manager.

Bug Fixes:

CRITICAL FIX - A bug that occurs when publishing more than 1 template from a single data source. Caused new publications to not detect forms if a single data source has more than 1 template published for it. Will now process correctly.
Fixed a bug where if there were more than a certain number of images in the processing queue and you opened that window it would cause the program to run out of memory. This was due to a poorly constructed SQL query.
Fixed the ordering of fields in the Form Page view for the captured data.
Fixed a bug with the automatic deskew of forms. Would only deskew data between the top quarter and bottom quarter of the page, and then cause an infinite loop if nothing was found between.
Fixed bug causing Asian, Arabic and other non-roman fonts not to appear in PDF exports.
Fixed bug in renaming of fields in the source data table.

Enhancements

Ability to have fieldnames and table names with any character set, including the use of spaces!
Ability to export and import all CSV files in UTF-8 format (international character support).
Added an enhancement to speed up the detection of forms that are scanned in any other orientation than the correct portrait rotation. Will now use last form's orientation to start the form detection process.
Toolbars are now scrollable if the screen becomes too small.
Set the default page size to either Letter (For US and Canada) or A4 for most other places. This will only be set when FormReturn first starts and no default paper size is already set.
Modified the name of OMR ID reconciliation to Key Field Reconciliation since both OMR Grids and Barcode Reader areas do the same thing.

CRITICAL UPDATE - November 27, 2009 - Version 1.0.6

New Features:

The very useful ability to drag and select in the editor, the page will scroll when you drag now so you can select elements further down the page.

Bug Fixes:

CRITICAL FIX - The form processor was crashing when too many images were in the unprocessed images list. The system would try to load all images into memory and subsequently run out of memory. Now fixed. If you had issues with forms not processing in the past, this version may fix those issues.
Embedded fonts wouldn't load correctly. Now fixed.

Enhancements

In prior versions, when making a rounded checkbox and export as a PDF file, the corners wouldn't look so good, particularly on smaller checkboxes. This is now fixed and you will no longer see incorrect corners.

November 25, 2009 - Version 1.0.5

New Features:

Ability to sort and name individual CSV export columns.
Ability to specify CSV export format.
Ability to save captured images that are stored in the database to disk.
Database automatically vacuums when the server is shutting down. This reclaims unused space in the database and reduces the file size of the database directory dramatically.
INTERNATIONALIZATION: The editor has some of the first stages of internationalization happening. It currently has translations for Spanish, Italian, German and French with more to come in the future. We will continue to internationalize all words in the program over the coming months.

Bug Fixes:

OMR ID Templates - Now hidden in the captured data forms listing so you don't accidently delete them.
Fixed TIFF image inversion problem - Some TIFF images had the colors inverted. Now fixed.
Fixed TIFF image stretched resolutions - Some TIFF images were displaying with a "stretched" appearance. Now fixed.
Fixed shutdown issue with FormReturn Server where it would ask to shutdown twice.

Enhancements

Mark aggregation rule now shows in checkbox panel area for one or more checkboxes.
Moved the captured results for form pages. More efficient use of space in this screen.
Made default buttons selected for most dialogs.
Database version checking - makes sure you don't run incompatible versions of the FormReturn client and Server together.

November 11, 2009 - Version 1.0.4

New Features:

Ability to apply checkbox preset styles to multiple checkbox selections. Useful for setting answer templates quickly - see screencasts.
Ability to modify multiple text areas at the same time. Useful for adjusting the fonts, size, style etc of more than one font quickly.

Bug Fixes:

SEMI-CRITICAL FIX: When FormReturn detected no mark for a mark area, the "and for everything else" rule didn't apply, it simply didn't record a score (or made a score of 0). This means if you put a negative score for the incorrect answer it wouldn't contribute toward the form score. This is now fixed.
OMR ID duplicate reconciliation keys issue - When a form used OMR ID - Reconcile or add new - any new reconciliation key record was duplciated. This was giving the appearance of random reconciliation key values in the source data view.
The Vera.ttf file might not have been installed properly in some versions of OS X. This has been fixed.
When adjusting the mark area in Windows, clicking the OK button will update the checkboxes in the editor - if you adjusted anything in the past you had to click on the element to see the changes. Now fixed.
The windows installer was giving Java version not found errors in 64 bit editions of Windows 7 and Vista for the non-JRE version of FormReturn. Now fixed.

Enhancements

The font system will notify you of fonts that can't be loaded. Please report any font that can't be loaded to bugs@ebstrada.com
Adding barcodes to the page that are too far right will now move the barcode left rather than resize it smaller. This will hopefully prevent people from making small barcodes (which make forms undetectable).
When editing mark area values (in the mark area properties dialog) - You no longer have to press tab or enter to make it accept the last value before clicking the OK button. Values are automatically adjusted every time you press a key.
When editing mark area values (in the mark area properties dialog) - You can now press a key such as "A" then tab then "B" then tab then "C" then tab etc so that when data is entered into the mark area values boxes, it is replaced - not appended. This makes it much faster/easier to enter multiple values and is more intuitive.

October 28, 2009 - Version 1.0.3

New Features:

When double clicking a FormReturn file such as a segment or form file in windows, it will automatically open in FormReturn.

Bug Fixes:

OSX INSTALLER FIX: Some Mac OS X computers had issues with the installer.. this is now fixed in this version. Would either say "zipexception" or not install the files properly.
SEMI-CRITICAL FIX: Some barcodes were not detecting properly. It would detect the barcode but not the bounds properly. This issue is fixed and the barcode detection has been enhanced.
CLOB/BLOB warning in the captured data / form screen fixed.
Segment height changing on zoom fixed.

Enhancements

New export status dialog. Allows you to see where the export is up to and abort it if necessary.
Faster loading of captured data manager tables.
Faster exporting of data.
Updated the default mark threshold to 40 and luminance to 200. This should suit most scanners better.

CRITICAL UPDATE - October 15, 2009 - Version 1.0.2

New Features:

Notifies user of incorrectly marked forms in the form preview.

Bug Fixes:

CRITCIAL FIX: Barcode detection now detecting all barcodes on a page. Previous bug caused some barcodes to miss detection. If forms were not detected in the past, please try rescanning them back into this version of FormReturn to see if they can be detected.
Fixed barcode size restoration: Barcode sizes were not restored when loading a saved Form. Now restores barcode sizes.
Fixed "add new segment" size dialog in the preferences panel. Was not appearing properly in Windows.
Fixed zoom issue in segment editor. When switching between the segment editor and its preview when zoomed in or out in the segment editor, resize issues were occuring.
Fixed segment recognition barcode scale size issues. Will now carry through any changes made to the scale.
Fixed perform deskew option in the recognition preview area. Will now carry this option through for the recognition preview testing.
Fixed recognition of the same barcode multiple times. Will now recognize the same barcode or side-by-side barcodes.

Enhancements

Mac OS X installer asks for admin privileges to install. This may fix issues where the user who installs FormReturn requires admin privileges. In the past has caused an "empty FormReturn folder" problem.

September 29, 2009 - Version 1.0.1

New Features:

Ability to edit the recognition settings after a publication has been published.
Ability to edit the mark aggregation rule after a publication has been published.
Measurements now available in pixels, millimeters, centimeters and inches for forms and segments.
Ability to download a published form template from the publication settings dialog.
New Letter size segments as well as A4. Click reset to default in preferences for default segment sizes and they will appear.

Bug Fixes:

Fixed preferences xml files to properly encode UTF-8. Caused FormReturn to stop loading on startup if you had a filename with a unicode encoded character in the preferences file. FormReturn will now properly handle unicode across client and server preferences.
If a fragment (checkbox) was too close to the border of a segment it would throw an exception and not recognize that checkbox. You can now have checkboxes touching the border and they will recognize.
If you published a form with empty source data records it may stop publishing and hang. Will now catch exception and notify you of empty records.
Fixed some ugly grey backgrounds around checkboxes in the GUI on windows.

Enhancements

Renamed "validate" for the mark aggregation rule to "test rule" to aid in clarity.
Faster loading of captured data view with removal of form count and page count.