Making Covoar More C++

[Joel Sherrill]

On Wed, Mar 24, 2021 at 2:42 PM Gedare Bloom <gedare at rtems.org> wrote:

> On Wed, Mar 24, 2021 at 1:35 PM Joel Sherrill <joel at rtems.org> wrote:
> >
> > Hi
> >
> > There has been a lot of talk about making covoar use more C++
> > features. It seems to be an issue on every patch. I almost
> > replied to Gedare's comment at the bottom of a patch
> > but decided it needed another thread:
> >
> > "I still struggle reviewing this codebase, in part because it is C+C++
> > (TM) and in part because I'm not so proficient in C++ to make concrete
> > recommendations how to write this better. I think, if the goal is
> > eventually to make this more C++ like code, then new code coming in
> > should aim to move the needle in that direction rather than continue
> > to propagate the old ways of doing."
> >
> Thanks for taking this on.
>
> > I personally do NOT want to see changes to C++ in one leaf class and
> > the other architectures not get the same changes. I would prefer to see
> > all these corrections and base changes in the same style with limited
> > changes to C-isms. I'm not opposed to the changes but let's take the
> > Target class one. There are multiple target classes. Changing one
> > independent of the others isn't a good idea.
> >
> This is reasonable to me.
>
> > I'd like to see us get a working baseline in and then do something like
> > sweep std::string through Target* as a single patch. This is easier to
> > test and review since it would only be C string to std::string. Perhaps
> > switch to C++ output a file at a time. Redo the report output. Etc.
> > Discrete chunks instead of piecemeal.
> >
> > Covoar has spent years broken and some is from changing working
> > things to do things "a better way" with no baseline to check against.
> > We need to get a baseline.
> >
> > Please. Let's get a working baseline and then approach this more
> > methodically. No one is going to suffer from seeing a C string a little
> > while longer. :)
> >
> I'm fine, as long as there is a plan in place and some clear
> directions. It would help to have tickets to organize the path
> forward.
>
> I'm willing to oblige continued use of C'ism for now, but I want to
> know the plan and maybe a deadline of sorts by which I can start to be
> picky again :) I don't like to be in limbo.
>

I'd love to have a deadline but I can't guarantee how long Alex can
work on it. But unless he gets pulled, he can pick on this for a while.

My guess is that C string to std::string is probably a good pass by
itself since method signatures may change.

There isn't much file input but that could be a pass by itself
along the way,

Then sweep output one file at a time leaving reporting for last as
a batch.

Do you see an order?

--joel

>
> > Thanks.
> >
> > --joel
> >
> >
> > _______________________________________________
> > devel mailing list
> > devel at rtems.org
> > http://lists.rtems.org/mailman/listinfo/devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20210324/417b8dc9/attachment-0001.html>