On 17/06/10 23:33, Andres P wrote:
On Thu, Jun 17, 2010 at 8:55 AM, Allan McRae<allan@archlinux.org> wrote:
On 17/06/10 22:44, Andres P wrote:
Tight variable scoping should avoid further regressions with new patches and variable overriding (see what ac5c2fd09 fixed).
Variable scoping had nothing to do with what ac5c2fd09 fixed. Maybe you mean another commit?
It had everything to do with it. Functions are big as it is, meaning the namespace is pretty cramped.
By chance there was a variable named changelog and install in scope.
Do remind that the original suggestion was to make that a seperate function, giving it its own set of local variables.
As I said, this had nothing to do with ac5c2fd09. Maybe you mean b8863622? Even then the issue was only in the case where multiple changelog or install files were specified in the PKGBUILD which the "$i=$(sed...)" syntax did not handle. Allan