[chbot] Lazy software developers please read... chocolate fish

[Andre Renaud]

> Yes I know all that, have done this regularly in gawk (because I found
> that easier than to do the same thing in sed). Never said you couldn't
> do it in sed, given a mental challenge, too much time, or both. But I'll
> give you all a chocolate fish at the next meeting if you show a sed
> program like
>     sed 's/xxx/yyy/'
> as suggested, that does do the job as asked, for a whole .c and a whole
> .h file. I'm even being generous and allow you multiple 's' commands,
> not as suggested.

Obviously it'd be nice to have a fully generic solution that covers
all of your code, no matter what style you write it in. However
assuming you're working on a single code base, and it's all been
written in a single style, then it often isn't too hard to match lines
that look like function prototypes. For instance, I tend to write all
function prototypes like this:
static int linked_list_move_to_end(linked_list_t *list, linked_node_t *node)

It starts in column 1 (or 0, depending on what editor you use), there
aren't line breaks between return types and function name, or between
function name and the first argument (although there may be some after
that, if there are a lot of arguments).

So for me, the following does a pretty good first pass at shoving a
template Doxygen header above every function:
sed 's,\(^[a-z].*(\),\n/**\n * Doxygen template\n */\n\1,g' FILE.c

Given the idea is simply to take some of the labour intensiveness out
of putting the documentation in, it doesn't have to be perfect as
you're obviously going to be going into each file by hand later on
anyway to put in sensible values, so if it misses the odd function, or
adds an entry in where it shouldn't be, removing those oddities is
less work than adding in the bulk of typical cases.

Regards,
Andre