README
1 List of bugs fixed. 2 3 1. ofs-rebuild: OFS value used to rebuild the record was incorrect. 4 Fixed August 19, 2014. Revised fix August 2018. 5 6 2. system-status: Instead of a floating-point division by 256, use 7 the wait(2) macros to create a reasonable exit status. Fixed March 12, 2016. 8 9 3. space: Use provided xisblank() function instead of ispace() for 10 matching [[:blank:]]. 11 12 4. a-format: Add POSIX standard %a and %A to supported formats. Check 13 at runtime that this format is available. 14 15 5. decr-NF: Decrementing NF did not change $0. This is a decades-old 16 bug. There are interactions with the old and new value of OFS as well. 17 Most of the fix came from the NetBSD awk. 18 19 6. string-conv: String conversions of scalars were sticky. Once a 20 conversion to string happened, even with OFMT, that value was used until 21 a new numeric value was assigned, even if OFMT differed from CONVFMT, 22 and also if CONVFMT changed. 23 24 7. unary-plus: Unary plus on a string constant returned the string. 25 Instead, it should convert the value to numeric and give that value. 26 27 8. concat-assign-same: Concatenation previously evaluated both sides of the 28 expression before doing its work, which, since assign() evaluates to the cell 29 being assigned to, meant that expressions like "print (a = 1) (a = 2)" would 30 print "22" rather than "12". 31 32 9. missing-precision: When using the format string "%*s", the precision 33 argument was used without checking if it was present first. 34 35 10. missing-precision: When using the format string "%*s", the precision 36 argument was used without checking if it was present first. 37 38 11. fmt-overflow: The buffer used for OFMT/CONVFMT conversions was written 39 to with sprintf(), which meant that some conversions could write past the 40 end. 41 42 12. numeric-subsep, numeric-fs, numeric-output-seps, numerics-rs: If SUBSEP, 43 FS, RS, OFS, or ORS were set to a numeric value, then their string values 44 wouldn't always be generated before being needed. 45 46 13. subsep-overflow: The length of SUBSEP needs to be rechecked after 47 calling execute(), in case SUBSEP itself has been changed. 48 49 14. split-fs-from-array: If the third argument to split() comes from the 50 array passed as the second argument, then split() would previously read 51 from the freed memory and possibly produce incorrect results (depending 52 on the system's malloc()/free() behaviour.) 53 54 15. getline-numeric: The `getline xx < file' syntax did not check if 55 values were numeric, in discordance from POSIX. Test case adapted from 56 one posted by Ben Bacarisse <ben.usenet (a] bsb.me.uk> in comp.lang.awk, 57 January 2019. 58
