Bonner Lab RUNOFF - Known bugs 1 Fall 82 John Clement Bonner Nuclear Labs Rice University P.O. Box 1892 Houston, Tx, 77251 April 29, 1982 The following is a list of known bugs in BONNER LAB RUNOFF. Each version fixed the bugs in the previous one. Some bugs were fixed in the course of making major changes, and may not appear in this list. Bugs in Fall 82 version ---- -- ---- -- ------- 1. Footnotes would not correctly justify and fill. This could have been worked around by using .NO FILL inside footnotes. 2. Footnotes would sometimes cause the first line of the next page to have incorrect fill and justification. 3. STYLE HEADERS wouldn't work in conjuction with NO HEADER, and TOP MARGIN 0. 4. .ELLIPSES would sometimes cause mistabulation. 5. .SUBPAGE at the beginning of the first page would cause RUNOFF to bomb. This is a nonsensical command sequence so nobody should have had problems with this one. 6. SUBSTITUTION failed unpredictably if a large number of substitutes were defined. 7. The /RI switch did not work at all properly. This could be worked around with a permanent left margin. 8. The last word of a section terminated by \& would not be underlined if the last word occured on the first word of an output line. 9. In .HEADER LEVEL command, if capitalization of the first character was desired and the first character was already in caps, the second char would be capitalized. The workaround was to not capitalize any section headers or use .STYLE HEADERS to disable automatic capitali- zation. 10. .LITERAL or .END LITERAL commands could bomb RUNOFF with an odd ad- dress trap. Also .LITERAL didn't turn off the flags properly if it worked. The use of .FLAGS ALL and .NO FLAGS ALL provided a partial workaround. 11. If overstriking preceded underlining on the same line, underlining by /UL:S or /UL:L would fail to underline correctly. /UL:B worked, however. 12. If both .NUMBER PAGE and .NUMBER SUBPAGE were used before a .SUBPAGE command, the page number would be wrong. 13. The number of spaces in a line could be messed up if the following conditions were met: .FILL .PERIOD were enabled and the end of the input line was terminated by a punctuation followed by a space. Since there is no need to have a space after a punctuation at the end of the line the workaround was to remove the final space. Bonner Lab RUNOFF - Known bugs 2 Spring 83 Bugs in Spring 83 version ---- -- ------ -- ------- 1. Equations imbedded inside text were not handled properly if the line had to be split. They are now properly moved to the next line. 2. Lines did not properly justify if they included equations. Specify- ing .NJ was the only solution in previous versions. 3. If .FILL was enabled and a printable escape sequence was braketed by a pair of spaces, the final space was removed. This no longer hap- pens if the HSP attribute is defined for the escape sequence. 4. If an escape sequence changed the pitch of the printer, improper un- derlining could result. If the pitch change was specified with the PSP and HSP attributes, then the underlining would be correct, other- wise the user had to use /UL:B instead of UL:L. 5. Module RNOIF was in the wrong overlay. It should have been in the same overlay as OPEN,CLOSE or an overlay that calls OPEN,CLOSE. This has been fixed. 6. RUNOFF would bomb when a .IF statement was not followed by the appropriate .ENDIF inside a "required file". 7. If hyphenation waenabled and escape sequences wee used, RUNOFF would occasionally hyphenate the output in strange places. The workaround was to .DHY during escape sequences. 8. Strange hyphenation could occur after tabs had been used. In general, hyphenation only worked properly for normal text. Now it should work properly for all text. Authohyphenation has been restricted to normal text, but user specified hyphenation can be used in any text except equations. 9. Roman numerals were incorrectly converted for 9,90-99,900-999. . . 10. .NUMBER LIST n started numbering at n+1 rather than n. .NUMBER LEVEL also worked incorrectly. The defaults have been modified for the .NUMBER commands to correspond exactly to DSR. 11. .LITERAL or .NO FLAGS ALL would turn off SPECIAL FLAGS permanently, never to turn on again. 12. The LCK attribute for escape sequences would not work correctly un- less .FLAGS ESCAPE command was used. 13. The /WA switch worked incorrectly when combined with /PA. RUNOFF would wait at the top of pages that were not being printed. Bugs in the Fall 83 version ---- -- --- ---- -- ------- 1. The subindex flag could not be used as a regular character after a .INDEX command, even when preceded by _. 2. Subindex terms with trailing or leading blanks were improperly in- dexed. 3. The index flag did not terminate an autoindex term. 4. .NUMBER LIST caused a fatal error in subsequent .LIST or .LIST ELE- MENT commands. 5. If .NUMBER CHAPTER and a .LAYOUT other than 0 was used, the last page of each chapter was improperly numbered. 6. .SUBPAGE, and .END SUBPAGE did not work correctly. The subpage letter was correct, but the page number was incorrect. Bonner Lab RUNOFF - Known bugs 3 Fall 83 7. .DEFINE COMMAND did not work properly if the defined command line in- cluded literal parameters. Unpredictable results occured when the command was used. 8. .IMMEDIATE TEST TEXT and .IMMEDIATE TEST PAGE incorrectly broke the input lines. 9. /-FF:n did not work correctly. n had to be in half lines, and if n=0, then an infinite loop resulted. 10. /PS:n also had to be in half lines. It is now in lines. 11. Occasionally when .JUSTIFY and .VARIABLE SPACING were in effect, the first line after the page header was improperly justified. 12. .NO FLAGS EQUATION caused an infinite loop if preceded by .FLAGS EQUATION. 13. .TEXT would sometimes lose sections of the input text if .SKIP or .BLANK occured at the end of a page. 14. When /UL:S was used, then /PA,/CH, and /AP would not work properly. 15. When /UL:L or /UL:S were used, and /PA,/CH or /AP were used with un- derlining in the title, the first page of output might have spurious underlining. 16. If there was an extra right brace (}) in an equation, RUNOFF died with an internal error instead of an illegal syntax error. 17. If /PA was used to suppress output pages before 256, page 256 would still be output. 18. Indirect command files did not work under IAS due to a bad overlay structure. 19. .ENABLE TOC and .DISABLE TOC would cause RUNOFF to bomb. This was fixed in module STYLE by changing BIC and BIS to BICB and BISB. 20. .DISABLE TOC disabled tabs, and .DISABLE TAB disabled TOC. 21. /U:L did not work properly under RT-11. 22. Null literals would sometimes cause dynamic memory overflow. Bugs in Spring 84 version ---- -- ------ -- ------- 1. .NO FLAGS ALL and .LITERAL would turn off flags that were in the page header. This caused incorrect titles, and subtitles. 2. The IASBLD.CMD file specified STACK=64, which is much too small. This line has been deleted. When RNO was built under IAS, a segment fault with stack overflow would result during execution. 3. A paragraph break that coincided with a page break would cause an ex- tra line to be inserted before the page number for layouts other than 0. 4. Substitutions that occured at a line break would cause spurious out- put and error messages. 5. Footnotes that were too long for the current page were incorrectly handled. 6. The widow and orphan preventions could cause spurious output when Footnotes were used. This is easily fixed by .SET PARAGRAPH ,,,0. Footnotes were split with intervening pages. 7. .LIST ELEMENT sometimes miscounted characters, resulting in incorrect output. 8. The HSP,VSP attributes for escape sequences would not accept zero, contrary to the documentation. Bonner Lab RUNOFF - Known bugs 4 Spring 84 9. The RT-11 version had a bug that caused an odd address trap on 11/73, but not on 11/23. The symbol TRCLN=17. in RNPRE.MAC was changed to TRCLN=18 to fix the problem. 10. The RT-11 version had a bug that caused dates for months after July to be incorrectly handled. It rolled over to January. This was, na- turally, discovered in August. 11. The .HALF SPACING command caused a skip, rather than setting the number of half spaced lines. 12. The letter representation of number was incorrect past Z of (26). 13. If the very first header level had no level specified, an error mes- sage was given. Now the first .HEADER LEVEL command is assumed to start at level 1. Bugs in Fall 84 Version ---- -- ---- -- ------- 1. .FILL mode was used inconsistently in headers, and page numbers. 2. .TAB PROPORTIONAL would give a fatal error if all columns were given a fixed width. .TAB PROPORTIONAL 3,10,10,10 ! Fails .TAB PROPORTIONAL 3,10,10 ! Works .TAB PROPORTIONAL 3,10,10,+1 ! Works 3. An .UNDERLINE command caused a punctuation (period, colon etc.) at the end of a line to insert 2 spaces. This messed up the justifica- tion. 4. A comma after a percent ",%" in a defined command was treated as a null parameter. It is now treated as a separator. Sometimes the parameter passing could malfunction if more than 1 level of substitu- tion was used. 5. .CHAPTER after .APPENDIX would still produce page numbers using the appendix number. Bugs in Spring 85 version 1. The RT version failed to output the last page if /E:N or /O:N was used. This could be partially worked around if the file was output edited so that it had an odd number of pages when printing odd pages, and an even number for printing only even pages. Bugs in Fall 85 version The following bugs were located by the RT Sig and forewarded to me by its charperson. Suggested fixes were also included. The author gra- tefully acknowledges the help he has been given. 1. .DEFINE ESCAPE,VARIABLE SPACING,SUBSCRIPT,SUPERSCRIPT will not handle nulls. A null inside an escape sequence would cause a fatal error. Nulls are now permissible. (submitted by: RT-SIG) 2. .LIST "c" set all subsequent lists at the same level to use the literal. (submitted by: Dan Eisner) 3. If symbol $DIABL was not defined the variable spacing was defined with garbage. (submitted by: Dan Eisner) 4. Escape sequences with non zero size caused bad underlining. Bonner Lab RUNOFF - Known bugs 5 Fall 85 5. Escape sequences are not handled properly in the native mode version inside header level commands, if pre and post fixes are not defined. 6. The default header level test page value was 14octal instead of 7. (submitted by RT-SIG) 7. The /RI switch under RSX was octal when it should have been decimal. (Courtesy of Jack Wenrick) 8. The following sequence would set the flags improperly: .NOFL UNDERLINE .FLAGS ACCEPT & .FLAGS UNDERLINE _ Essentially the accept flag will not work. .NOFL UNDERLINE .FLAGS ACCEPT & .FLAGS UNDERLINE _ This should give an error message, but improperly treats & as an un- derscore flag. Submitted by RT-SIG 9. When RNO is asked to print 2 documents in a row some of the trailing section of the first document may appear at the beginning of the second. Submitted by RT-SIG 10. STYLE HEADERS unlike some other commands reinitializes all values which are not specified. This is in conformance with DSR. The do- cumentation has been changed to reflect this. Submitted by RT-SIG 11. The RT version may swallow the last few lines of output when /P op- tion is used. Submitted by RT-SIG 12. Escape sequences with VSP attribute can mess up the line count per page. Submitted by RT-SIG 13. The command .CENTRE was not recognized, and .END CENTRE was not available. Submitted by RT-SIG 14. The last line is not processed if it is not terminated by a , in the RT version. Sometimes RNO randomly swallows text in the RT version due to improper end of buffer finding. There were a number of potential problems in the RT version which have been fixed. Submitted by RT-SIG 15. The REQUIRE BINARY command rejects the line immediately following the command. It worked if the next line was a comment. Bugs in Fall 86 version 1. It was not possible to center justify an equation with the equation number right justified. This only showed up when tab stops were used with equations. The correct behavior (pre Fall 86) was restored. Known Bugs Not Currently Solved ----- ---- --- --------- ------ 1. .RESET SUBSTITUTE after .DEFINE NUMBER PAGE could cause a fatal er- ror. This is prevented by a .FLUSH command befor the .RESET. Bonner Lab RUNOFF - Known bugs 6 Current Bugs 2. .RESET INDEX with deferred pages containing index entries pending could cause a fatal error. This is prevented by a .FLUSH command befor the .RESET. 3. Any .RESET SUBSTITUTE inside a defined command will cause fatal er- rors. 4. An accepted character just after the last char. in a line will not be properly handled if the line has no spaces, breaks, or hyphenation points. 5. It is possible that there is still an interaction between footnotes, widow preventions, and text sections. The first thing to try is turning off widow prevention, using .SET PARAGRAPH ,,,0. There is currently no known problem, but since these features are fairly com- plex, a bug could be lurking. NOTE If you find any bugs I would appreciate getting a copy of both the .RNO and .DOC files containing these bugs. If you could isolate the bug and demonstrate it in just a few lines of input I would appre- ciate the effort. I will try to fix bad bugs or at least create a workaround. I do not guarantee anything.