GNU Awk 3.1.2 has problems when parsing multiline records. see "Steps to Reproduce" for details. Reproducible: Always Steps to Reproduce: $ od -c data 0000000 a \n \n b \n \n c \n 0000010 $ cat pr.awk BEGIN { RS = "" } { printf("%d: %s\n", NR, $0) } END { print NR } $ cat data | awk -f pr.awk 1: a 2: b 2 Actual Results: the last record (#3, containing 'c') is not printed Expected Results: record #3 should be printed... i sent a bug report regarding this problem to bug-gawk@gnu.org. stepan kasal sent me a patch which fixes the problem. it's available from http://www.math.cas.cz/~kasal/gawk/gawk-3.1.2-rsnull2d.patch.gz i'm not sure wether any of the core utilities depend on multiline records or not, so i file this report with severity "major".
arnold robbins - the maintainer and developer of gnu awk - said (regarding the multiline records problem): "This bug is only in 3.1.2, it is a bug. For now, the solution is to move back to 3.1.1. I am working on 3.1.3 and hope to release it "soon", which will fix the problem."
Az: think we should back down the stable version and wait for 3.1.3?
yes.
marked unstable. Will wait for 3.1.3 for a new stable gawk.
And now it's marked stable again, still containing the bug... ???
Apply patch or mask again ? Maybe do a new revision of 3.1.1 afterwards ?
I'd mask it again, because of Arnold Robbins statement. But I'm not sure if no other packages depend on functionality added to Awk between 3.1.1 and 3.1.2...
Nope, latest revision 3.1.1 have filefuncs.so in.