IRIX 6.4 Applications 1997 August

home *** CD-ROM | disk | FTP | other *** search

/ IRIX 6.4 Applications 1997 August / SGI IRIX 6.4 Applications 1997 August.iso / relnotes / performer_eoe / ch5.z / ch5

Wrap

Text File | 1997-08-01 | 13.3 KB | 331 lines

- 1 - 5. _B_u_g_s__F_i_x_e_d__b_y__P_e_r_f_o_r_m_e_r__2_._0_._4_/_2_._1_._2_,__p_a_t_c_h__S_G_0_0_0_1_6_9_6 The IRIS Performer 2.0.4 and Patch SG0001696 libraries are binary compatible with IRIS Performer 2.0-2.0.3 and includes compatibility subsystems to upgrate the DSOs shipped with IRIS Performer 2.1. There were some problems in the 2.0 and 2.1 libraries that were fixed for the 2.0.4 execution environment. This section list the bugs fixed in each incremental release of IRIS Performer 2.0.1, 2.0.2, 2.0.3, 2.0.4 and in the 2.1.1 and 2.1.2 compatibility libraries. 5.1 _I_R_I_S _P_e_r_f_o_r_m_e_r _2._0 _P_r_o_b_l_e_m_s _f_i_x_e_d _i_n _2._0._1 _a_n_d _2._0._2 _a_n_d _2._1 +o _O_p_e_n_G_L _F_o_g: OpenGL exponential fog in IRIS Performer 2.0 did not match the IRIS GL fog and has been corrected. (bug #338726) +o _O_p_e_n_G_L _r_e_q_u_e_s_t _f_o_r _T_A_G _c_l_e_a_r _o_n _I_n_f_i_n_i_t_e_R_e_a_l_i_t_y _c_a_u_s_e_d _O_p_e_n_G_L _e_r_r_o_r_s (bug #342228) +o _O_p_e_n_G_L _d_e_f_a_u_l_t _M_I_P_m_a_p_p_e_d _T_e_x_t_u_r_e _f_i_l_t_e_r: Graphics platforms that should be using MIPMAP_TRILINEAR as the default minification filter (such as RealityEngine and IMPACT) were actually using MIPMAP_LINEAR in 2.0. The default minification and MIPmap filter has been fixed. (bug #392320) +o _D_e_t_a_i_l _T_e_x_t_u_r_e _S_p_l_i_n_e_s: Several bugs with the specification of detail splines in IRIS Performer 2.0 have been fixed. The default OpenGL splines now match the IRIS GL defaults. Specification of clamped detail splines for IRIS GL operation have been fixed. (bug #339589) +o _O_p_e_n_G_L _D_e_t_a_i_l _T_e_x_t_u_r_e _o_n _I_R_I_X _6._2: OpenGL detail texture through IRIS Performer 2.0 was not functional when running under 6.2 and has been fixed. (bug #392259) +o _O_p_e_n_G_L _T_e_x_t_u_r_e _L_o_a_d_i_n_g _o_n _R_e_a_l_i_t_y_E_n_g_i_n_e: In IRIS Performer 2.0 running under OpenGL on RealityEngine graphics, texture had to be explicitly enabled for a texture to be successfully loaded. This is now handled internally by IRIS Performer. +o _p_f_d_L_o_a_d_F_i_l_e__f_l_t: The FLT loader under IRIS Performer 2.0 64bit operation was core dumping and this has been fixed. (bug #392262) - 2 - +o _O_r_d_e_r _o_f _p_f_B_u_f_f_e_r _c_o_m_m_a_n_d_s _e_x_e_c_u_t_e_d _b_y _p_f_M_e_r_g_e_B_u_f_f_e_r() _i_s _r_e_v_e_r_s_e_d _i_n _2._0: IRIS Performer 2.0 could core dump after pfMergeBuffer() if child nodes in the asynchronous process were created before their parents. (bug #347746) +o _p_f_D_r_a_w_B_i_n _c_o_r_e _d_u_m_p: pfDrawBin() would core dump if called on an empty bin. (bug #344198) +o _I_R_I_S _G_L _w_i_n_d_o_w_s _h_a_d _n_o _s_t_e_r_e_o _b_u_f_f_e_r_s: pure IRIS GL windows would not be configured with stereo buffers with pfWinFBConfigAtrrs() even if the PFFB_STEREO token was present in the attribute array. (bug #344181) +o _I_R_I_S _G_L _w_i_n_d_o_w _q_u_e_r_i_e_s _w_e_r_e _b_r_o_k_e_n: some of the window config queries for pfQueryWin() would always return 0 for pure IRIS GL windows. (bug #344181) +o Resizing of pfPipeWindows when multi-processed and using X windows (IRIS GLX or OpenGL/X) when an alternate framebuffer configuration window is selected (such as the fill statistics window in OpenGL/X perfly) has been fixed. In IRIS Performer 2.0 this could channel viewports to be confused when the alternate framebuffer configuration window is de-selected (such as disabling the fill statistics in perfly when running with X windows). (bug #392262) +o _p_f_A_l_p_h_a_F_u_n_c _h_a_d _n_o _e_f_f_e_c_t _i_f _P_F_S_T_A_T_E__E_N_L_I_G_H_T_I_N_G _i_s _b_e_i_n_g _o_v_e_r_r_i_d_d_e_n (bug #319457) +o _p_f_M_a_t_r_i_x::_g_e_t_O_r_t_h_o_C_o_o_r_d: Performer2.0 it gave bad roll when pitch was +-90. (bug #334615) +o _C++ _p_e_r_f_l_y _e_r_r_o_r: In the C++ perfly and related programs, on multipipe machines, all windows get opened on pipe 0. The fix is swapping lines 419 and 420 in perf/sample/apps/C++/common/generic.C. (bug #335105) +o Intersections with PFGS_POLYS could cause core dumps in Performer2.0. (bug #346482) +o _M_e_m_o_r_y _c_o_r_r_u_p_t_i_o_n _b_y _e_x_t_e_n_d_i_n_g _p_f_S_t_r_i_n_g_s _b_y _1 _c_h_a_r_a_c_t_e_r: (bug #346502) +o _B_a_d _m_a_l_l_o_c _f_o_r _n_a_m_e_s _o_f _p_f_F_o_n_t_s: this could potentially cause a core dump in Performer2.0 +o _p_f_T_e_x_t_u_r_e_s _c_o_u_l_d _n_o_t _s_u_b_l_o_a_d _M_I_P_m_a_p _l_e_v_e_l_s (bug #334345) - 3 - +o _p_f_G_e_t_T_e_x_B_o_r_d_e_r_C_o_l_o_r did not work in Performer2.0. (bug #332384) +o _R_e_f_l_e_c_t_i_o_n _m_a_p_p_i_n_g _w_i_t_h _t_h_e _I_n_v_e_n_t_o_r _l_o_a_d_e_r: files loaded through libpfiv could not properly use reflection mapping in Performer2.0. (bug #324961) 5.2 _I_R_I_S _P_e_r_f_o_r_m_e_r _2._0 _a_n_d _2._0._1 _P_r_o_b_l_e_m_s _f_i_x_e_d _i_n _2._0._2 _a_n_d _2._1 +o _N_3_2 _c_o_r_e _d_u_m_p _i_n _i_n __p_f_D_i_r_t_C_h_e_c_k _a_n_d __p_f_L_i_g_h_t_D_i_r_t_C_h_e_c_k: Due to an N32 compiler bug, N32 applications with more than 256 textures or lights could core dump in _pfDirtCheck or _pfLightDirtCheck. A work-around for this problem is in 2.1 and 2.0.2. (bug #399553) +o _p_f_T_e_x_t_u_r_e_s _w_i_t_h _s_o_u_r_c_e _o_f _P_F_T_E_X__S_O_U_R_C_E__V_I_D_E_O did not work under OpenGL operation. This has been fixed in 2.0.2 and in 2.1. (bug #396077) +o _I_n_t_e_r_n_a_l _f_o_r_m_a_t _f_o_r _1 _a_n_d _2 _c_o_m_p_o_n_e_n_t _t_e_x_t_u_r_e_s: defaulted to 12 bits for the intensity component which causes inefficient memory usage on IMPACT. +o _p_f_T_e_x_t_u_r_e _r_e_f_e_r_e_n_c_e _c_o_u_n_t_s _i_n_c_r_e_m_e_n_t_e_d _a_f_t_e_r _p_f_L_o_a_d_T_e_x() _i_n _l_i_b_p_f _a_p_p_s: pfTextures that are loaded with pfLoadTex() or pfSubloadTex{Level} could get their reference count artificially incremented in libpf applications. (bug #392273) +o _p_f_D_e_l_e_t_e _w_i_t_h _f_o_r_k_e_d _D_B_A_S_E _p_r_o_c_e_s_s: asynchronous pfDelete of data allocated off the heap and not in shared memory could cause core dump. (bug #392274) +o _p_f_F_l_a_t_t_e_n _o_n _i_n_d_e_x_e_d _p_f_G_e_o_S_e_t_s _c_o_u_l_d _c_a_u_s_e _c_o_r_e _d_u_m_p. (bug #365204) +o _p_f_L_P_o_i_n_t_S_t_a_t_e_s _u_s_i_n_g _t_e_x_t_u_r_e_s _f_o_r _l_i_g_h_t _p_o_i_n_t _s_h_a_p_e generated an incorrect default texture for the shape of the light. (bug #392317) +o _p_f_M_a_t_r_i_x::_m_a_k_e_V_e_c_R_o_t_V_e_c() _g_a_v_e _a _b_a_d _m_a_t_r_i_x _o_n _o_p_p_o_s_i_t_e _v_e_c_t_o_r_s: When the input vectors were pointing in opposite directions, a matrix was returned that represented a uniform scale of -1, which is not a rotation. This has been fixed in 2.0.2 and 2.1. (bug #402541) +o _C++ _p_e_r_f_l_y _h_a_d _r_a_n_d_o_m_l_y _b_a_d _f_l_i_g_h_t _m_o_d_e_l _b_e_h_a_v_i_o_r: The C++ perfly example source code had the call to pfSync() - 4 - commented out in main.C and this caused problems with the flight model when intersections are run in a separate process (bug #402637). 5.3 _I_R_I_S _P_e_r_f_o_r_m_e_r _2._0-_2._0._2 _a_n_d _2._1 _P_r_o_b_l_e_m_s _f_i_x_e_d _i_n _2._0._3/_2._1._1 +o _S_u_b_l_o_a_d_i_n_g _o_f _d_e_t_a_i_l _t_e_x_t_u_r_e_s: The subloading of detail textures (pfLoadTex() of textures of format PFTEX_SUBLOAD_FORMAT) and is fixed in 2.0.3 and in the 2.1.1 compatibility DSOs. +o _C_o_m_p_i_l_a_t_i_o_n _o_f _p_f_G_e_o_S_e_t_s _i_n_t_o _O_p_e_n_G_L _D_i_s_p_l_a_y _l_i_s_t_s _n_o_t _o_p_t_i_m_a_l _f_o_r _i_R: The compilation of pfGeoSets into OpenGL Display lists was using joint compilation and execution which resulted in the GL display lists not being simultaneously downloaded into the graphics pipeline and could then cause an unexpected performance hit the first time an object was drawn to get it downloaded. +o _p_f_Q_u_e_r_y_S_y_s _o_f _P_F_Q_S_Y_S__M_A_X__M_S__S_A_M_P_L_E_S _w_a_s _b_r_o_k_e_n _o_n _i_R: f2pfQuerySys of PFQSYS_MAX_MS_SAMPLES incorrectly returned 0 for small and extra large pixel depths on InfiniteReality graphics subsystems. +o _T_K_O (_t_r_i_p_l_e _k_e_y_b_o_a_r_d _o_p_t_i_o_n _f_o_r _m_u_l_t_i_p_i_p_e _s_y_s_t_e_m_s): tko was broken after the initial release of Performer 2.0 and is now fixed in 2.0.3/2.1.1. 5.4 _I_R_I_S__P_e_r_f_o_r_m_e_r__2_._1__P_r_o_b_l_e_m_s__f_i_x_e_d__i_n__2_._0_._3_/_2_._1_._1 +o _p_f_T_e_x_t_u_r_e _r_e_f_e_r_e_n_c_e _c_o_u_n_t_s _i_n_c_r_e_m_e_n_t_e_d _a_f_t_e_r _p_f_L_o_a_d_T_e_x() _i_n _l_i_b_p_f _a_p_p_s: pfTextures that are loaded with pfLoadTex() or pfSubloadTex{Level} could get their reference count artificially incremented in libpf applications. (bug #392273) +o The array of MIPmaps in pfTextures was not correctly inernally reference counted and could result in a very small and slow memory leak for applications paging pfTextures with explicitly specified MIPmap levels. +o _p_f_b _l_o_a_d_e_r _f_o_r_c_i_n_g _p_f_G_e_o_S_e_t _c_o_m_p_i_l_a_t_i_o_n: The pfb loader was forcing on GL compilation of pfGeoSets if the file had been written with pfGeoSets in the compiled state. - 5 - +o _P_b_u_f_f_e_r_s _c_a_u_s_e_d _e_x_i_t _w_i_t_h _X _e_r_r_o_r: If a pfWindow or pfPipeWindow was given an OpenGL pbuffer as its drawable it would cause Performer to exit with an X error. Now, there is very minimal support for pbuffers: pbuffers can be provided as a drawable with pf[P]WinWSDrawable if in addition the visual and corresponding GL context are also provided. 5.5 _I_R_I_S _P_e_r_f_o_r_m_e_r _2._0._3 _a_n_d _2._1._1 _P_r_o_b_l_e_m_s _f_i_x_e_d _i_n _2._0._4/_2._1._2 +o _A_s_s_i_g_n_i_n_g _W_S_D_r_a_w_a_b_l_e_s _t_o _p_f{_P_i_p_e}_W_i_n_d_o_w_s _w_i_t_h_o_u_t _a _v_i_s_u_a_l: There was a regression in 2.0.3/2.1.1 where if a pfWSDrawable (such as taken from a motif rendering area) was assigned to a pfWindow or pfPipeWindow and the matching X visual was not also provided, the window would fail to open. This regression is ifxed in 2.0.4 and 2.1.2.