Metagraphics MetaWINDOW Revision History
Versions 4.1A through 4.3D

Fixed in Version 4.3D

• TextWidth() with stroked text and non align left would not return the correct width the first time it was called.
• VESA support now available for Rational 4GW
• FrameRect with solid wide lines was not taking fill optimization correctly, was drawing as 4 vectors (introduced in 4.2)
• WriteImage() of monochrome source to multi-plane offscreen bitmaps did not color expand correctly.
• Stroked text and TextAlign(x,y) the y alignment was only correct at paths 0 and 180
• Oval pen, (round really), flat and round caps, round and bevel joins are now implemented.
• Frame Ovals, Arcs now dash.
• Rational 4GW libs now do lock code areas used for interrupt service routines when using Virtual Memory Manager (VMM).
• Added VESA modes for 32K and 64K color.
• Added VESA modes for 1280x1024 256 color Note: ATIGUP must use BIOS (palette #1) to set palette in 1280 mode
• Patterns greater than 8x8 on ATI Ultra in 1280x1024 would reset hardware clip to 1024 wide.
• SetPort() with lower left origin would not bump the pen position up when converting user pen location to global
• Negative clip coordinates would reverse min with max on 386

Fixed in Version 4.3C

• XlateImage() from 1 bit/pix would mess up if image alignment + x pixel position exceeded 16.
• Stroked text did not draw end cap pixel on last vector of each char (introduced in v4.2C).

Fixed in Version 4.3B

• Pascal had coordAbs and coordRel reversed for FillPolygon()
• PolyMarker in Pascal did not work
• LineTo after PolySegment put an extra line down left side of screen

Fixed in Version 4.3A

• S3 WriteImage when clipped did not output correct data to bitmap.
• XlateImage with large images in 386 generated wrong images.
• Stroked text would cause a GP fault when Y went -, would not draw the end point of line segments when it should (random behavior, reading bad pointer for end cap status).
• Genoa 6000 board 1024x768x16 had banking problems
• VESA drivers had problems with certain chipsets (ie Cirrus Logic)
• Attempting an InitBitmap for XMS when no XMS services are installed would cause a system hang upon CloseBitmap().
• Calling CloseBitmap() when InitBitmap() failed could free a driver prematurely.
• Multiple EMS bitmaps in some cases would not bank in the correct rowtable banks when switching between them.
• Using Pascal, FloodFill/BoundaryFill did not work.
• Using Pascal, HardCopy would not work if passed a device name (ie PRN, LPT#, FILE.PRN, etc).
• S3 boards with 512K would not fill correctly when using a pattern.
• InitBitmap() for cDISK was not setting the DOS Disk Transfer Area for Find First File operations, resulting in stack corruption.
• Some Trident boards did not work right, especially in the 640x480 256 color mode.
• XMS bitmaps with an odd number of bytes per raster and an odd number of rasters per swap buffer would fail when accessed.
• Pascal GrafFree was not freeing 8 bytes per call
• Pascal cHalftone was not defined
• Pascal XlateImage was incorrectly declared as a procedure, s/b function
• Pascal InitMouse was incorrectly declared as a procedure, s/b function
• Pascal GrafAlloc was not checking for nil pointers returned
• Pascal GrafAlloc and GrafFree were not public

Fixed in Version 4.2B

• XlateImage on 386 did not return 32 bits worth of size.
• For 386, blits 7 bits wide from mono source to 8 bit had vertical bars. Usually noticed with proportional fonts.
• Using any S3 board, bitmapped fonts (usually proportional) in 8 bit mode, have black bars & scrunched chars (VGB-30.FNT).
• CopyBlit() from 8514 or S3 in protected mode to an offscreen bitmap caused a GP fault.
• CloseBitmap() with a partially allocated bitmap (InitBitmap failed) freed a bogus pointer
• ClipRect after a ClipRect(NULL) didn't clip.
• QueryGraphics was erroneously returning 0 (Device located) instead of -1 (Device status cannot be determined) for some cases.

Fixed in Version 4.2A

• SetPenState() state didn't restore text rasterop.
• RoundRect with height of 2 would hang.
• The label "fixed" conflicted with the C++ IOSTREAM class, changed to type fxdpnt.
• Stroked text with TextSlant() had bugs
• TextFace() now operational
• PortOrigin(lowerLeft) caused some routines to update the port pen location incorrectly (one pixel up). Affected routines: DrawString(), QueryCursor(), PeekEvent(), KeyEvent(), PtOnOval().
• Phar Lap 386 libs now do lock code areas used for interrupt service routines when using Virtual Memory Managers (VMM).
• HardCopy() to a file that already existed mucked the file up
• HardCopy() caused a GP fault on 386 in some cases
• XMS and DISK bitmaps would allocate the wrong size swap buffer, depending on the type of malloc used.
• ScrollRect offscreen would set the pen color to 1, and the back color to 0. In 16/256 color modes, things would draw blue. (only in 4.1D)
• Screen to screen blits (ScrollRect, CopyBlit) did not work for VESA 640x480, 800x600, & 1Kx768 256 color modes. (only in 4.1D)

Fixed in Version 4.1D

• HardCopy() printed garbage in 4.1C Premium for 386
• FrameArc() overdrew end point when angle = 270.0
• SetPort(NULL) did not work
• ZoomBlit() did not clip correctly for all port positions
• ZoomBlit() zoomimg down would mess up the dominate color computation on source colors of white
• PortBitmap() on XMS or DISK bitmaps caused last cache buffer to not be flushed out, leaving muck when CopyBlit to screen.
• PortBitmap() would 'fix up' the port size if port larger than bitmap, this causes a bogus MovePortTo if origin lower left. This 'fixup' was bogus anyway and was removed.
• ScrollRect used current rasterop for blit, now uses replace always
• ScrollRect on monochrome would invert the blit colors if pen color not = 1 and back color not = 0
• CopyBlit crashed if source rect is completely or partly off the screen bitmap.
• CopyBlit did not clip correctly for all cases of port positions
• Multi plane (offscreen) bitmaps GetPixel() was broken (affected ZoomBlit)
• DupRegion didn't update region structure pointer segments
• GetPixel could return bad value if Tracking on and mouse moved (also affected flood and boundary fill)
• Arcs (paint or frame) sometimes crashed
• ClipRect() when in virtual coords used bad address to convert rect
• FrameRect() with wide pen dashed lines did not dash
• Vir2GblPt saved point to bad address
• KeyEvent() or PeekEvent() returned bad mouse x,y for keyboard events when in local or virtual coordinates
• Virtual Coordinate transformations did not check for divide overflow correclty, causing divide overflow for coordinates that would have been between (+ or -) 32767 to 65536
• GrafPool() only worked if called prior to InitGraphics(), now works dynamically.
• Ovals 2 pixels high would crash
• ClipRect( NULL ) did not disable clipping
• When text totaly clipped via region on 8514, would crash
• Using Turbo Pascal, DrawChar did not work
• Any S3 board using the new 86C924 chip, did not work.

Fixed in Version 4.1C

• InitMouse() returned bogus values. Always.
• CloseBitmap() did not deallocate XMS memory.
• EmptyRect() return value was backwards
• PolyLine() with wide pen and origin lower left would draw one pixel too low.
• Drawing text with a clip region where nothing should be drawn would hang.
• ZoomBlit() was broken from 4.1A, where is used to work.
• Lower left origin had some bugs.
• CopyBlit() and ZoomBlit() did not convert clip rect correctly when using virtual coordinates.
• PtOnxx and PtInxx did not take the WIDTH and HEIGHT parameters in global pixels, so virtual sizes would sometimes scale to 0.
• Pen position was not updated when MovePortTo() was called, resulting in weird StringWidth() results.
• ScrollRect() would crash if preceeded only by a WriteImage().
• MetQuery.c wouldn't accept your video mode command line arg unless it was the last arg on the command line.

Fixed in Version 4.1B

• XMS bitmap support added
• DrawText() was left out of the lib.
• TextWidth() didn't work properly.
• ScrollRect() may not scroll from correct bitmap.
• ScrollRect() deltas not correct under local or virtual.
• Point in/on hit rectangle wasn't centered properly.
• PtInFlood(), PtInBoundary() added.
• ReadPCX() now reads from current bitmap, instead of default bitmap
• Interlaced bitmaps (CGA, Hercules, etc.) corrupted memory by writing 4 bytes beyond the allocated rowtable memory.
• CenterRect() didn't match manual definition.
• PtOnxx() drew frame when pen size = 0,0.
• PtIn/Onxx() with hit size = 1,1 did not test any pixel.
• Fixed FillArc() and FrameArc() to correct intermittent hangs under Borland C when drawing special cases of even height ovals with end angles at 360.0.
• InitGraphics() would hang if using Phar Lap 286 and fonts (or drivers) were not in the current directory.
• TrackCursor(True) would sync the cursor to the input device, now syncs the input device to the current cursor position.
• ProtectRect() would hang if called before ShowCursor().
• SetPenState() corrupted memory when it called PortBitmap().
• AlignPattern with y of 1 crashes.
• SetPixel() with raster ops other that replace, or with patterns other than 0 or 1 would not draw.
• XlateImage on 386 blows up

====== History starts with version 4.1A ======

MetaWINDOW Revision History - Versions 4.4A through 5.0B

Return to Metagraphics Home Page