Jamie Nguyen
dysco****@gmail*****
Sat Dec 18 02:15:51 JST 2010
TOMOYO
ForkTetsuo Handa wrote:
> Now, 'old_{path,head,tail,number,address}_pattern' lines are interpreted just
> like 'domain' and 'acl' lines, and evaluation stops at first chunk which
> reached the last line of that chunk (i.e. one of
> 'new_{path,head,tail,number,address}_pattern' lines).
>
> But I feel that 'old_{path,head,tail,number,address}_pattern' and
> 'new_{path,head,tail,number,address}_pattern' are misnamed.
> These keywords convey that the former refers old value and the latter refers
> new value, but these keywords do not convey that the former is interpreted as
> precondition and the latter is interpreted as chunk terminator.
> Something like
>
> acl[1].equals file
> acl[2].equals read
> path_pattern /tmp/cc\*
> rewrite @TMP_CC_FILES
>
> might be better than
>
> acl[1].equals file
> acl[2].equals read
> old_path_pattern /tmp/cc\*
> new_path_pattern @TMP_CC_FILES
Yes, something like rewrite seems better.
> OK. I modified to use separate lines.
> Initial content of /etc/ccs/tools/patternize.conf as of revision 4208 is
>
> # Files on proc filesystem.
> old_path_pattern proc:/self/task/\$/fdinfo/\$
> new_path_pattern proc:/self/task/\$/fdinfo/\$
> old_path_pattern proc:/self/task/\$/fd/\$
> new_path_pattern proc:/self/task/\$/fd/\$
> old_head_pattern proc:/self/task/\$/
> new_head_pattern proc:/self/task/\$/
Do you require new_* in these cases, as new is same as old? Is it
still possible to not include new_pattern? If not, I suppose that
"rewrite" without a second argument could terminate the chunk without
having to repeat the pattern.
Kind regards