On Fri, Oct 23, 2009 at 3:09 PM, Parrot <[email protected]> wrote:
> #1137: [PATCH] fix codingstd/c_indent.t wrt switch/case labels
> ------------------------------+---------------------------------------------
> Reporter: plobsing | Owner: jkeenan
> Type: patch | Status: assigned
> Priority: minor | Milestone:
> Component: coding_standards | Version: 1.6.0
> Severity: low | Keywords:
> Lang: | Patch:
> Platform: |
> ------------------------------+---------------------------------------------
> Changes (by jkeenan):
>
> * status: new => assigned
>
>
> Comment:
>
> I tried this patch out in trunk and ran `prove`. The output is attached.
>
> As it stands, I think the patch creates more problems than it solves. It
> flags as errors many code chunks that we probably don't think of as bad C
> code formatting. For example, it flags 3 errors in this code from
> ''src/io/sockets_unix.c'':
> {{{
> 171 case EINPROGRESS:
> 172 goto AGAIN;
> 173 case EISCONN:
> 174 return 0;
> 175 default:
> 176 return -1;
> 177 }
> }}}
> Here we have 3 lines ending in a colon outdented by 4 characters rather
> than 2. They pass the trunk version of ''c_indent.t'' but not the patched
> version. I suspect that the problem is that, intuitively, we only want
> '''LABELs''' to be outdented by 2; we want all other indentation to be by
> 4. I think this was what plobsing was trying to get at. However, it's
> very difficult to embody this precisely in code because it's difficult to
> say that one line ending in a colon is a LABEL whereas another, as in a
> switch statement, is not.
>
> Other thoughts/opinions?
>
> Thank you very much.
>
> kid51
Looks like making
if ($line =~ /:$/) {
slightly smarter would be sufficient; say, optional leading whitespace
followed by valid label characters ending with a colon - that should
be enough to stop it from catching 'case FOO:'.
--
Will "Coke" Coleda
_______________________________________________
parrot-tickets mailing list
[email protected]
http://lists.parrot.org/mailman/listinfo/parrot-tickets
