home *** CD-ROM | disk | FTP | other *** search
-
-
-
- - 1 -
-
-
-
- 5. _K_n_o_w_n__P_r_o_b_l_e_m_s__a_n_d__W_o_r_k_a_r_o_u_n_d_s
-
- 5.1 _I_R_I_X__6_._2
-
- For OpenGL issues on all platforms, please consult the
- OpenGL man pages. Known bugs and machine-dependencies have
- been documented there, so that they are readily available
- during development and debugging.
-
- 5.1.1 _I_n_f_i_n_i_t_e_R_e_a_l_i_t_y
-
- +o ``Old-style'' stereo is not supported currently.
- ``New-style'' (quad-buffered) stereo is supported. We
- expect to handle old-style stereo in a future release.
-
- +o X11 line performance is lower than we would like. This
- will be fixed in a future release.
-
- +o X11 polygons drawn with a clip mask may not be clipped
- correctly. This will be fixed in a future release.
-
- +o ``Gang swap'' (multipipe buffer-swap synchronization)
- is not supported. It will be supported in a future
- release.
-
- +o /usr/gfx/gfxinfo does not report per-channel
- information about the constituency of channels and
- video. This will be corrected in a future release.
-
- +o IrisGL 3-component lookups in a 4-component texture
- color lookup table (tlut) do not duplicate the blue
- component for use as an alpha index. Instead, an alpha
- of 255 is assumed.
-
- +o The X server must be running in order to use the
- ``setmon'' command. This will be changed in the
- future.
-
- +o X rendering using the IncludeInferiors subwindow mode
- will not work correctly across windows of the same
- depth but in different layers, or windows of the same
- depth but having different framebuffer pixel
- representations.
-
- 5.1.2 _R_e_a_l_i_t_y_E_n_g_i_n_e_,__R_e_a_l_i_t_y_E_n_g_i_n_e_2_,__a_n_d__V_T_X
- Note: please see the OpenGL man pages for further
- information about OpenGL limitations on these systems.
-
- +o lrectwrite() with images wider than 2048 pixels will
- not draw anything.
-
-
-
-
-
-
-
-
-
-
-
-
- - 2 -
-
-
-
- 5.1.3 _V_G_X__a_n_d__V_G_X_T
-
- +o When readdisplay() is called with the RD_FREEZE hint in
- a mixed-model application, the call never returns and
- the X server no longer responds to user input. To
- workaround, remove the RD_FREEZE hint and surround the
- readdisplay() call with ``XGrabServer(XtDisplay(w));''
- and ``XUngrabServer(XtDisplay(w));''.
-
- 5.1.4 _M_a_x__I_m_p_a_c_t_,__H_i_g_h__I_m_p_a_c_t_,__a_n_d__S_o_l_i_d__I_m_p_a_c_t
-
- +o only 4444 packed formats work for tex image code
- (349693)
-
- +o gfx driver should report genlock achieved (315082)
-
- +o setmon manpage incomplete (320870)
-
- +o glGetTexLevelParameteriv returns bogus *_SIZE_EXT
- (324856)
-
- +o glBlendColorEXT() state: use it or loseit (326111)
-
- +o Framebuffer: getting junk values:Impact-Prime (323969)
-
- +o IGLOO: holes in shared polygon edges (318483)
-
- +o GLX_BUFFER_SIZE is not the sum of r,g,b,a on Impact and
- RE (309671)
-
- +o Texture objects shared between OpenGL contexts do not
- work correctly (311507). This affects Inventor
- applications such as WebSpace Author and Iris
- Annotator, which may drop textures from their displays
- on Impact systems. An environment variable,
- IV_NO_TEXTURE_OBJECT, may be set (to any value) to work
- around this problem. Also, IrisGL on Impact must work
- around this problem in a way which slows down some
- texturing operations. This will be fixed in a future
- release.
-
- +o The background may change color when an IrisGL program
- is run as the background. This includes the X and SGI
- logos from the standard desktops backgrounds. (349308)
-
- +o The texture magnification filter
- GL_LINEAR_DETAIL_COLOR_SGIS is not supported on Impact
- systems (303738).
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- - 3 -
-
-
-
- 5.2 _I_R_I_X__5_._3
-
- 5.2.1 _R_e_a_l_i_t_y_E_n_g_i_n_e_,__R_e_a_l_i_t_y_E_n_g_i_n_e_2_,__a_n_d__V_T_X
-
- +o On Crimson, accumulation buffer and texture mapping
- with subtex loads fail to work.
-
- +o swapinterval has a timing hole where if the swap is
- requested 1 vertical retrace before the actual swap
- time the swap is delayed.
-
- +o glRasterPos with ColorIndex does not work.
-
- +o OpenGL cannot bind a context alternately to a window
- and then a pixmap. The result is that the rendering
- will be wrong.
-
- +o PROXY does not work for the OpenGL Texture extension.
-
- +o OpenGL Depth Buffer reads don't work in multisampling
- mode.
-
- +o TexSubImage2D is broken for indirect rendering. It
- makes the implicit assumption that one has previously
- called TexImage2D with a NULL image. The protocol does
- not handle this case.
-
- +o GL_SRC_ALPHA_SATURATE does not work with
- GL_REVERSE_SUBTRACT_FUNC.
-
- +o XCopyArea copies incorrectly from a window to a pixmap.
-
- +o ABGR DrawPixels is broken for a zoom factor other than
- 1.0.
-
- +o ReadPixels of alpha for visuals that don't have alpha
- will yield incorrect values for alpha for the last
- pixel of each span read. The expected values is 1.0
- but the value will be 0.0.
-
- +o ReadPixels for the depth buffer is broken.
-
- +o OpenGL does not support rendering into any underlay
- visuals.
-
- +o With multi-pass rendering, if some passes use texturing
- and others do not, the Z values may not be exactly the
- same.
-
- +o When RGBsize(8) is set, z-buffer readbacks might return
- incorrect data.
-
-
-
-
-
-
-
-
-
-
-
- - 4 -
-
-
-
- +o Pixel transfers larger than 2048 x 2048 do not work.
-
- +o Texture coordinates from the t3f command are not
- correctly transformed by the 3x3 texture matrix.
-
-
- 5.2.2 _R_e_a_l_i_t_y_E_n_g_i_n_e__w_i_t_h__M_u_l_t_i_-_C_h_a_n_n_e_l__O_p_t_i_o_n
-
- +o When installing Multi-Channel Option hardware on a
- system equipped with RealityEngine graphics where IRIX
- 5.1 is already installed, you must manually specify the
- VME address of the Multi-Channel Option if it is not
- the default. See the MCO Installation Guide for
- details on the _v_s_2_a_d_d_r_s_e_t program.
-
- Note that if the Multi-Channel Option hardware is
- already installed, installing IRIX 5.1 automatically
- specifies the default address.
-
- +o Now that cursor tracking works on most formats, it
- should be noted that it still does not work in dual
- field-sequential RGB mode.
-
- +o The 3@1025x768 format requires that the RealityEngine
- graphics organization be manually configured to support
- 10-bit RGB without alpha. To do so, edit the file
- /_u_s_r/_b_i_n/_X_1_1/_X. The file appears as:
-
- ####!!!! ////bbbbiiiinnnn////sssshhhh
- ####
- #### ////uuuussssrrrr////bbbbiiiinnnn////XXXX11111111////XXXX
- #### iiiinnnniiiittttiiiiaaaalllliiiizzzzeeee ggggrrrraaaapppphhhhiiiiccccssss ddddeeeevvvviiiicccceeee aaaannnndddd ssssttttaaaarrrrtttt XXXX WWWWiiiinnnnddddoooowwww SSSSyyyysssstttteeeemmmm
- #### $$$$RRRReeeevvvviiiissssiiiioooonnnn:::: 1111....22225555 $$$$
- ####
- ggggllllGGGGaaaammmmmmmmaaaaFFFFiiiilllleeee====////eeeettttcccc////ccccoooonnnnffffiiiigggg////ssssyyyysssstttteeeemmmm....ggggllllGGGGaaaammmmmmmmaaaaVVVVaaaallll
- ggggllllGGGGaaaammmmmmmmaaaaDDDDeeeeffffaaaauuuulllltttt====""""1111....7777""""
-
- iiiiffff ////eeeettttcccc////cccchhhhkkkkccccoooonnnnffffiiiigggg wwwwiiiinnnnddddoooowwwwssssyyyysssstttteeeemmmm;;;; tttthhhheeeennnn
- iiiiffff [[[[ ----xxxx ////uuuussssrrrr////ggggffffxxxx////ggggffffxxxxiiiinnnniiiitttt ]]]];;;; tttthhhheeeennnn
- ////uuuussssrrrr////ggggffffxxxx////ggggffffxxxxiiiinnnniiiitttt
- iiiiffff [[[[ ----rrrr $$$$ggggllllGGGGaaaammmmmmmmaaaaFFFFiiiilllleeee ----aaaa ----ssss $$$$ggggllllGGGGaaaammmmmmmmaaaaFFFFiiiilllleeee ]]]] ;;;; tttthhhheeeennnn
-
-
- Modify the file to include the following statements
- between line 11 and line 12:
-
- PPPPIIIIXXXXDDDDEEEENNNNSSSS====PPPPIIIIXXXX33332222____RRRRGGGGBBBB11110000
- eeeexxxxppppoooorrrrtttt PPPPIIIIXXXXDDDDEEEENNNNSSSS
-
- The file should appear as:
-
-
-
-
-
-
-
-
-
-
-
-
- - 5 -
-
-
-
- ####
- #### ////uuuussssrrrr////bbbbiiiinnnn////XXXX11111111////XXXX
- #### iiiinnnniiiittttiiiiaaaalllliiiizzzzeeee ggggrrrraaaapppphhhhiiiiccccssss ddddeeeevvvviiiicccceeee aaaannnndddd ssssttttaaaarrrrtttt XXXX WWWWiiiinnnnddddoooowwww SSSSyyyysssstttteeeemmmm
- #### $$$$RRRReeeevvvviiiissssiiiioooonnnn:::: 1111....22225555 $$$$
- ####
- ggggllllGGGGaaaammmmmmmmaaaaFFFFiiiilllleeee====////eeeettttcccc////ccccoooonnnnffffiiiigggg////ssssyyyysssstttteeeemmmm....ggggllllGGGGaaaammmmmmmmaaaaVVVVaaaallll
- ggggllllGGGGaaaammmmmmmmaaaaDDDDeeeeffffaaaauuuulllltttt====""""1111....7777""""
-
- iiiiffff ////eeeettttcccc////cccchhhhkkkkccccoooonnnnffffiiiigggg wwwwiiiinnnnddddoooowwwwssssyyyysssstttteeeemmmm;;;; tttthhhheeeennnn
- iiiiffff [[[[ ----xxxx ////uuuussssrrrr////ggggffffxxxx////ggggffffxxxxiiiinnnniiiitttt ]]]];;;; tttthhhheeeennnn
- PPPPIIIIXXXXDDDDEEEENNNNSSSS====PPPPIIIIXXXX33332222____RRRRGGGGBBBB11110000
- eeeexxxxppppoooorrrrtttt PPPPIIIIXXXXDDDDEEEENNNNSSSS
- ////uuuussssrrrr////ggggffffxxxx////ggggffffxxxxiiiinnnniiiitttt
- iiiiffff [[[[ $$$$???? ----eeeeqqqq 0000 ----aaaa ----xxxx ////uuuussssrrrr////bbbbiiiinnnn////XXXX11111111////XXXXssssggggiiii ]]]];;;; tttthhhheeeennnn
-
- 5.2.3 _V_G_X__a_n_d__V_G_X_T
-
- +o Some concave polygons may be drawn as convex. This
- only occurs when the vertices of the polygon are very
- close together.
-
- +o readdisplay not capturing underlay planes.
-
- +o Two-sided lighting might render incorrectly on old-
- style rectangles.
-
- +o lsetdepth malfunctions when using GL/X mixed-model.
-
- 5.2.4 _O_t_h_e_r__H_i_g_h_-_E_n_d__C_o_n_f_i_g_u_r_a_t_i_o_n_s
-
- +o On R4K / GT as soon as one vertex of a display list is
- clipped the entire display list is clipped.
-
- +o The 4Dgifts program _z_r_g_b._c demonstrates that the IP4GT
- drops polygons sometimes when the polygons are near
- clipped.
-
- 5.2.5 _X_S_,__X_Z_,__E_l_a_n_,__a_n_d__E_x_t_r_e_m_e
-
- +o When rendering with concave(TRUE), very small concave
- and convex polygons (less than 1 pixel in size in
- screen space) fail to draw sometimes. Users should
- avoid setting concave(TRUE) when drawing data sets that
- do not contain concave polygons. The only solution for
- such very small concave polygons is for you the user to
- tessellate the concave polygon into triangles.
-
-
-
-
-
-
-
-
-
-
-
-
-
-