1 
  2 09/21/87  graphics
  3 Known errors in the current release of graphics.
  4 #         Associated TR's
  5 Description
  6 
  7 99  phx15957
  8 Can't use calcomp_915 module since ntape_ got unsupported.
  9 
 10 98  phx12939
 11 setup_graphics when used twice without a remove_graphics in between,
 12 attempts to do the remove_graphics for the user.  However, if -online
 13 is used on the second setup_grahics command, it will say that it is in
 14 error.
 15 
 16 97  phx12908
 17 calls get_to_text_mode too early to set rawo mode, thus sometimes not
 18 going to ^rawo mode when done.  This happens for any GSP that returns
 19 any characters from its close entrypoint.
 20 
 21 96  phx12390
 22 Uses graphic output buffer to store modes returned from GSP.  If mode
 23 string exceeds Message_size, it can get shipped to terminal by mistake.
 24 
 25 95  phx12256
 26 Uses enter_graphic_mode entry to set plot number on tape.  This results
 27 in resetting the plot number erroneously if the user should use
 28 changemode on his graphic switch.  The open entry or a similar
 29 mechanism should be used.
 30 
 31 94  phx12179
 32 DSB-chaining code always references through null pointer when creating
 33 second and subsequent DSB's.
 34 
 35 93  phx11126
 36 Attach description and modes for graphic_dim_ not well documented in
 37 AS40.
 38 
 39 92  phx11825
 40 remove_graphics assumes all online switches go to tty_i/o, even when
 41 set up elsewhere via sg -online -to xxx.
 42 
 43 91
 44 Passes unrecognized control orders except when they are "sub-orders"
 45 of the "io_call" order.  It should pass these, too.
 46 
 47 90  phx11334
 48 Use of the "999." feature of ccs_$symbol and ccs_$number after an
 49 intervening call to most other entrypoints in ccs_ will produce
 50 incorrect placement/sizing of the output.
 51 
 52 89  phx10008
 53 Calcomp tape format only accommodates 63 characters per text string,
 54 but calcomp_915_ ignores this limitation.
 55 
 56 88  phx09736
 57 Module calcomp_915_dim_ leaves attach table in half-baked state if
 58 attach through it fails.
 59 
 60 87  phx09637
 61 Masking of event calls, necessary to keep asynchronous output (notably
 62 interactive messages) from appearing in mid-picture, has side-effects
 63 including interfering with tape I/O.  There is no general solution to
 64 this problem, and the only workable write-around is non-trivial.
 65 
 66 86
 67 graphics_editor (show in particular) cannot handle max-sized
 68 lists/arrays.  When the current limit is removed, the whole strategy
 69 will have to be reworked in any case.
 70 
 71 85  phx09633
 72 The graphic_dim_ never invokes GDT's at their "Prepare_for_graphics"
 73 entries.  (EXL only)
 74 
 75 84  MPRF06514
 76 Return conversion errors occur in gm_$examine_list if supplied array is
 77 too small.
 78 
 79 83
 80 graphic_dim_ allows event-calls (e.g., interactive messages) to occur
 81 at the most inopportune moments.  (EXL only)
 82 
 83 82
 84 Misdeclaration of argument in graphic_dim_order causes set_tty command
 85 to complain about undefined order requests.
 86 
 87 80  MPRF06412
 88 Compatibility feature "-offline" to setup_graphics gives various
 89 errors, process terminations, and so on.
 90 
 91 79  phx08622
 92 Run units and setup_graphics don't mix.  The perprocess-static switch
 93 needs to be set in the bindfile for bound_graphics_system_.
 94 (Temporarily set 12/09/80 in both >exl and >unb via
 95 perprocess_static_sw_on command.)
 96 
 97 78  phx08462
 98 String extents arguments returned by graphic_chars_$long and $long_tb
 99 fail to account for character scale factors supplied in call.
100 
101 77
102 The new graphic_code_util_ uses a stack temporary reserved for
103 pl1_operators_.  (EXL only)
104 
105 76
106 When detaching a graphic switch, graphic_dim_ first closes out the GDT
107 and then tries to tell it to go into text mode.  This bug was caused by
108 the fix to bug 66.  (EXL only)
109 
110 75
111 Bug in tektronix_40XX_ causes it to forget to zero an error code which
112 it ignores anyway when it is writing to a file instead of to the
113 terminal.
114 
115 74  phx09014
116 graphic_compiler_ puts out spurious shifts before text elements in
117 lists (not arrays).
118 
119 73
120 A bug introduced with the implementation of suggestion 44 causes the
121 graphic_compiler_ to mess up coordinates of single elements and
122 elements of lists.  (EXL only)
123 
124 72
125 graphic_dim_ faults in modes entry if graphics attached to a file
126 through a GDT.
127 
128 71  phx07892
129 lsm_$set_blk does not complain about attempts to grow lists or arrays
130 past their max-- instead, it copies the entire list over again in the
131 WGS and doesn't grow it at all.  This results in the WGS filling up
132 rather quickly with useless dead lists when using things like ccs_.
133 
134 70  phx07462
135 ge replay will print very small coordinate values in "e" format,
136 even though when they are re-input, it cannot understand them.
137 
138 69  phx07461
139 gr_print_ reverses blinking and sensitivity when interpreting graphics
140 code.
141 
142 68  phx07431 phx07842
143 If a mode or mapping occurs as the last element of an array, the
144 graphic_compiler_ will lock up solid.
145 
146 67  phx07453
147 Scaling in graphic_editor, when given no arguments, defaults to "0 1
148 1" instead of "1 1 1".  Looks like an incomplete fix to bug 34.
149 
150 66  phx07285
151 If any characters are output by a GSP's close entry, the graphic_dim_
152 will leave the process in rawo mode.
153 
154 65  phx06411
155 plot_ will attempt to take log of a negative value when performing auto
156 scaling of data, when all data consists of identical negative numbers.
157 
158 64  phx06160
159 Never really draws points or setpoints.
160 
161 63
162 Sometimes plot_ will come out with something silly like 4.2345e-9
163 instead of zero for an axis value.
164 
165 62
166 Signalling of arbitrary condition in graphic_dim_ should be changed to
167 call to sub_err_.
168 
169 61  phx05706
170 Requests to ccs_$plot to reset the origin fail if the current graphic
171 position already happens to be at the designated point.
172 
173 60
174 The graphic_decompiler_ signals things instead of using sub_err_.
175 
176 59
177 list_pgs_contents initiates the pgs directly instead of using lsm_; if
178 the segment is not a real pgs it will fault or otherwise fail.  (EXL
179 only)
180 
181 58
182 lsm_fs_$init_seg never checks the segment it is initiating to see if it
183 even resembles an lsm_ segment.
184 
185 57
186 Even though users have >exl>gs in their search rules before >unb, use
187 of parts of the graphics system for which no experimental copy exists
188 will sometimes cause the process (via the referencing_dir rule) to use
189 bits and pieces of >unb versions of the graphics system.  The most
190 common symptom is for one version to create a WGS and for the other to
191 swear it doesn't exist.  The only workable fix for this is to keep an
192 entire complement of the executable bound segments in EXL whether they
193 have been changed or not.  (EXL only)
194 
195 56
196 If graphic_element_length_ is used incorrectly by a user (the MGS
197 validates better than this) it signals "illegal_graphic_effector".
198 This should at least be a call to sub_err_.
199 
200 55  phx05493 phx06847
201 Entry gm_$examine_color fails because it uses an unset pointer.
202 
203 54  phx05492
204 Input arrays of more than 4095 elements cause graphic_decompiler_ to
205 garbage its stack and fail.  This bug may be finessed instead by a
206 timely implementation of suggestion 52.
207 
208 53
209 Support of color terminals is not possible until the redundancy
210 inherent in defining both color and intensity effectors is dealt with;
211 intensity is physically a function of color as color is defined in the
212 MGS.  The fix for this design deficiency is to de-define the intensity
213 effector altogether, which must await implementation of planned
214 improvement 46.
215 
216 52
217 The MR8.0 MGS refuses to operate on version 6 PGS's having a symbol in
218 slot 0 of the hashtable.  Attempts result in spurious and
219 uncircumventable "Node out of bounds" messages.
220 
221 51
222 Tektronix devices at 9600 baud sometimes draw crooked lines.  This is
223 due to an undocumented restriction on the speed at which these devices
224 can draw lines.  Manufacturer's suggested solution is to disable
225 character transmission optimization software for terminals at this
226 speed or above.
227 
228 50
229 When automatically upgrading a PGS to the version 7 format,
230 gm_$use_file (really lsm_fs_$pull) creates a symbol hashtable one word
231 too long (and thus of nonprime size).
232 
233 40
234 Tektronix GSP does not clip to the size of the screen.  Vectors that
235 cross the edges of the screen become badly distorted.