On Mon, Apr 6, 2009 at 11:57 PM, Mike Walter <[email protected]> wrote:
> BTW, wouldn't best coding practices imply a change of the 'CMS' stages to > 'COMMAND' stages (with all uppercase arguments)? > Pity of someone has their own 'LISTDIR' or 'QUERY' exec on an accessed > disk. Yes, that's a VERY bad practice - but why expose your code to their > very bad practice? As Kris points out, there's more to the "cms" versus "command" stages than the search for an EXEC with the same name. So the Religion of Address Command does not come without sacrifice. I too played with invoking CMDCALL within the "cms" stage to avoid suppressing messages for unexpected errors, but the lack of documentation made me avoid it so far. One should also realize that with multiple unrelated pipeline segments invoking "cms" or "command" you have no way to predict the order in which they run. Sometimes that's fine, but if you want predictable results you may need to run them through a funnel. That also makes it much easier to debug your pipeline, and I consider it an aspect of personal hygiene not to have all kind of side effects leaking out of the pipeline... Sir Rob the Plumber
