On Thu, Mar 07, 2002 at 10:03 AM, Jean-Luc Leger wrote:
> > Larry McVoy <[email protected]> writes:
> > > bk prs -hrv2.5.0.. | while read x
>
> by the way, shouldn't it be "$x" in the second line ?
> or am I missing something ?
>
> JL
man bash
...
read [-ers] [-t timeout] [-a aname] [-p prompt] [-n
nchars] [-d delim] [name ...]
^^^^
Nope, no "$" on the variable name in this context. The reference
is to the variable's identifier rather than its value.
Ed Vance
On Thu, Mar 07, 2002 at 10:56:08AM -0800, Ed Vance wrote:
> On Thu, Mar 07, 2002 at 10:03 AM, Jean-Luc Leger wrote:
> > > Larry McVoy <[email protected]> writes:
> > > > bk prs -hrv2.5.0.. | while read x
> >
> > by the way, shouldn't it be "$x" in the second line ?
> > or am I missing something ?
> >
> > JL
>
> man bash
> ...
> read [-ers] [-t timeout] [-a aname] [-p prompt] [-n
> nchars] [-d delim] [name ...]
We're getting well into programming 101 and I should just shut up, but
here:
cat > mkpatches <<EOF
#!/bin/sh
bk prs -hr"$1".. | while read x
do bk export -tpatch -r$x > /tmp/patches/patch-$x
done
EOF
Yes, I believe you need cat and a Bourne compatible shell for the <<EOF
stuff and we really don't need a listing of all the shells in which this
will and will not work.
The point, which seems to be completely lost in the "I can program in
shell better than you" discussion, is that it is trivial to export
changes from BK as patches. Given that the world turns on patches
today, that should be the end of the discussion and insofar as I am
concerned, it is the end of the discussion.
--
---
Larry McVoy lm at bitmover.com http://www.bitmover.com/lm
On Thu, 7 Mar 2002, Ed Vance wrote:
> On Thu, Mar 07, 2002 at 10:03 AM, Jean-Luc Leger wrote:
> > > Larry McVoy <[email protected]> writes:
> > > > bk prs -hrv2.5.0.. | while read x
> >
> > by the way, shouldn't it be "$x" in the second line ?
> > or am I missing something ?
> >
> > JL
>
> man bash
> ...
> read [-ers] [-t timeout] [-a aname] [-p prompt] [-n
> nchars] [-d delim] [name ...]
> ^^^^
> Nope, no "$" on the variable name in this context. The reference
> is to the variable's identifier rather than its value.
He said _second_ line. Larry got $i there. In any case, for that
one
bk prs -hrv2.5.0.. | sed -e "s!.*!<body of the loop>!" | sh
would be simpler and more compact (with & instead of $i).
He doesn't have metacharcters in there, so it's safe...