Known bugs

The bug numbers refer to our internal bug database (and no, we don't have tens of thousands of MAT bugs - the bug database covers a huge number of projects).

• "Save rich" in the UI is always assumed to succeed, since the system asks the browser to save the document and there's no possibility of feedback. This means that if the document has been changed, the system will believe that the changes have now been saved, whether or not the user completes the save action in the browser. Future undo/redo steps will not prompt the user to confirm that unsaved changes can be undone.
• Bug 15811: Using borders as CSS properties for annotation displays creates numerous problems in the UI: the boundaries will appear on the smallest subsegment (tokens, usually) instead of around the intended phrase, and the stacking margins in the annotation and comparison views will be miscalculated.
  
• Bug 16227: rerunning the experiment engine, except to regenerate scores, seems to be a good idea which is poorly implemented.
• Bug 23906: The Java service API only permits JSON files to be imported into workspaces.
• Bug 35858: In some limited circumstances, if you change the value of an annotation attribute in the UI and then click "OK" in the annotation popup editor, the UI will reconfigure itself in such a way that it moves the "OK" button out from under your mouse and fails to dismiss the dialog, due to the order of blur and click methods in the UI library.
• Bug 36574: The xml-inline writer preserves set and list attribute values by concatenating them using a comma, but it doesn't escape the comma in string values.
• Bug 36827: Because CSS is case-insensitive, annotation labels which differ only in case will not be distinguishable in the UI presentation.
• Autotagging in the UI intentionally fails to copy annotation-valued attributes. We do this because anchored events, when autotagged, will almost always want their own argument fillers. We recognize that there are situations where we'd want to retain these attributes, and other situations in which we'd want to fail to copy certain string-valued attributes. We hope to address these issues in a future release.
• Autotagging is available only in the annotation popup in the annotation palette, not the popup in the annotation table.
• The interaction of comparison/reconciliation with spanless annotations without display properties is currently undetermined, and will probably break.
• If you swipe past the end of a document, you won't get an annotation popup because the mouse-up event isn't processed by the document pane.
• If your annotation window is just wide enough to hold all the annotation tabs you've opened, and one of the tabs holds an unmodified document, and you modify that document, the tab list will wrap, but won't redisplay the top of the window below it correctly. You can recover from this by resizing your browser window or by hiding and unhiding one of your tabs.
• If any characters in the document signal are not in Unicode plane 0 (otherwise known as the Basic Multilingual Plane), it's possible that offset counting will be inconsistent. The reason for this is that both Java and JavaScript default to counting UTF-16 code elements, and so without doing anything special, characters above the BMP have length 2. It's a gory tale, which you can read more about here.
• If you build a default model while MATWeb is running, and the model did not previously exist, MATWeb will not find it. You must restart the server. You do not have to reload your annotation tool in the browser.