On 07/26/2014 07:06 PM, Bernhard Voelker wrote: > seq(1) could not determine the width of large values toward INF > values. Therefore, specifying -w is useless in such a case. > > * src/seq.c (main): Output an error diagnostic when the -w option > is specified together with non-finite values for FIRST or LAST. > * tests/misc/seq.pl: Add tests. > * NEWS (Changes in behavior): Mention the change.
We already restrict -w to be incompat with --format. However that's an invariant aspect of seq invocation. One might have a script that took the end point as a parameter, which would now fail. I've a slight preference for the existing behavior of ignoring -w in this case, like we do for hex inputs for example. thanks, Pádraig.
