New bug reports
Setting font in Helium editor causes fractions to display incorrectly.
- 18 Jul 2008
Helium editor: Chess board does not clear possible move rectangles after move.
Dazzle editor: backspace at the start of a label crashes Proxima.
Focus up/down does not work in line containing a fraction (Helium editor).
Changes in the presentation may cause Proxima to crash when computing the new focus.
Newly generated presentation id counter may produce double id's.
When the counter passes maxint, it starts at 0 again. When this is about to happen, the idps should be garbage collected, for example, if there are n ---
idp's, we replace them by [0..n], and start counting at n+1.
IDP's for lists are not handled right.
The presentationAG just adds 100 to each list. Instead, we could specify the amount of IDP's required for each separator, and multiply it by the length of the list.
(StringP ... `addPopupItems` ...) does not work. Workaround: surround the string by a row or column.
Unconstrained stretching presentations do not fill the entire window.
possible bug: growing presentation may not incrementally update dependent stretching presentations.
Ctrl-d (structural delete) does not work.
maybe we should use ctrl-x and check for a structural selection
Slide parser in Helium editor does not work.
deleting in a line using the presentation selection does not clear the old presentation of the line.
e.g. in "abcdefg", deleting "a[bc]defg" results in "adefgfg"
The nodes for Structural, Parsing, and Locator in the diff tree were always self-clean, even if their child was not. Hence, updatedRectArr did not handle shrinking well.
Solution: In diffArr, the nodes for structural, parsing and locator (possibly a chain of these) now take their self-dirty status from their child.
Formatters in presentations prevent stretching elements to fill the entire window.
col [poly [(0,0),(1,1)], formatter [ text "bla" ]]
does not fill the window.
minHeight for Formatter was 0 whereas it should have been the height of the unfolded formatter.
Formatters are clipped when there is not enough width available.
minWidth is now the maximum minwidth of the children.
Arrangemer incrementality does not work well with window resize.
Resize now invalidates entire presentation in prunePresentation.
Initial presentation is rendered as a black rectangle.
check on empty rendering was done on old rendering instead of new.
ctrl-v crashes Proxima.
Copy did not handle formatters well. A clip from a formatter is now represented by a RowP.
on parsing, focus is lost.
The scanner now stores the focus in the extra state on the presentation level. It is restored during the Layout phase.
Placing cursor right on some lines crashes Proxima.
(../proxima/src/Layout/Scanner.hs:147:33-75: Irrefutable pattern failed for pattern (left, focusedChar : right) )
Cannot navigate past empty row or column (but no crash).
Presentation navigation (left & right) should now be able to handle any presentation.
When typing text, tokens all get the same presentation ID, leading to strange behavior.
Before reusing the presentation ID in the scanChars, we check whether the whitespaceMap already has this token, and if so, create a new one.
Delete in vertex labels does not work.
have been extended with cases for
Generator needs dummy declaration.
All types that are defined must be used on the right-hand side somewhere, or code generation fails. Hence a
declaration is sometimes needed in
Dummies are no longer necessary
Adding a vertex in graph gives a stack overflow.
had a black hole
On parse error, the focus is lost.
Focus is stored in whitespace map and restored automatically.
When moving over an empty line, the presentation focus is not cleared from the screen.
Focus near and in structural presentations is sometimes invisible and sometimes not cleared.
Selection containing (part of) a power exp is incorrect. Focus is not displayed correctly, and delete does not work.
All fixed: 19-4-2008.
Focus is no longer part of the rendering, but is a separate rendering in the rendering level. It is added to the rendering bitmap on every paint operation. For the old solution, we needed to mark the part of the arrangement with the old and new focus, causing complexities and bugs, and making it impossible for the size of the focus to be bigger than the size of the focused element. In the new solution, these problems are solved.
Scanner does not handle columns well.
Elements with no children are converted to in XML, causing the parser to fail.
GHC 6.8.2 compilation fails on GADT pattern.
Presentation.Prestypes: GADT pattern match in non-rigid context for `TokenP'
- 19 May 2008
Fixed: 8-7-2008: added type sigs to all functions that pattern match on TokenP