Re: [Groff] Overview, Sept. 2014

[Ralph Corderoy]

Hi,

Ingo wrote:
> But look at the actual C++ we are talking about here.  There are
> neither templates nor rocket science in there.

Yes, it's nowhere near as bad as modern C++.  

> That can be combined with writing unit tests if one feels so inclined,
> which is another desideratum.

Ingo, have you any opinion on a test framework?  I wasn't thinking so
much unit tests, but higher level std{out,err} checks.  Being a GNU
project, there's obviously DejaGnu, but I've long thought the world's
moved on from needing Expect capability and writing Tcl.  Ad hoc shell
seems the obvious choice, building up a library of shell functions for
common tests, but you might have better ideas.

> Or to accept the code base as is, start with integrating own code
> adhering to the existing style, and get intimate with an area of the
> code during that process

That's my favourite.  Its conventions might be different, given its age,
but I think it's pretty consistent.  Enthusiasm for anything other than
trivial changes in style would soon wear thin if bugs were being
introduced.  And given the limited scope of the code base, is wrapping
every man and his dog in a setter()/getter() really needed just because
it seems the norm in the C++/Java world?  (The code isn't really C++,
more C with Classes.)

Regarding knowing whom to ask for help;  ask the list.  There any many
lurkers here and one might pipe up on their pet topic.  It's also a way
of publicising where work's being cogitated or done.  I think the list
for conversation and the issue tracker as its more accessible cache for
on-going things should work well.

Finally, I know modern MUAs, e.g. Gmail, turn a thousand lines of quoted
email into a single folded line, but please delete it from the reply
otherwise the rest of us have to page through it checking that no new
text has been interspersed, the way emails should be written.  ;-)

Cheers, Ralph.