On 10/19/15 3:45 AM, grishale...@gmail.com wrote:
> When using ${parameter:offset:length} expansion, the parser doesn't account
> for all the ways that a colon can appear in "offset".
>
> For example:
>
> $ echo ${PARAM:${OFFSET:-0}}
> -bash: ${OFFSET: bad substitution
>
> or:
>
> # declare -A A=([a:b]=0)
> # echo ${PARAM:${A[a:b]}}
> -bash: ${A[a: bad substitution
>
> It seems that the issue is in the use of skiparith in
> verify_substring_values. Switching to to skip_to_delim removes this issue.
> The ${parameter/pattern/string} expansion uses skip_to_delim to search for
> the second '/' so I suspect it's use is appropriate here as well.
It's not quite that simple. The reason skiparith() is used is that
arithmetic expressions may contain a colon (test?exp1:exp2), so skiparith
skips over one colon for each `?'. I will have to figure out a way to
continue to support this syntax for backwards compatibility and allow for
colons in other contexts.
Chet
--
``The lyf so short, the craft so long to lerne.'' - Chaucer
``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, ITS, CWRU c...@case.edu http://cnswww.cns.cwru.edu/~chet/
