[Geany-Devel] Let's move to C++98 - Re: Lets move to C99

[Dimitar Zhekov]

On Mon, 26 Aug 2013 16:07:13 +0100
Nick Treleaven <nick.treleaven at btinternet.com> wrote:

> I now think gradually using a (quite heavily restricted) subset of C++98 
> for *some* source files might be better than moving to C99 [...]
> 
> Reasons:
> 
> * RAII - this is a pretty essential feature for safe resource 
> management.

This works for global/static and auto classes. We can wrap a gchar *s
= g_strdup_printf() in a class, but I'd rather not... and you want to
ban "class".

> * References can be useful in making code more readable vs pointers.

But using both heavily (as GLib/GTK+ required pointers) will make it
harder to read.

> * Developer enjoyment, productivity & gaining more experience with C++. 
> (This might be an incentive for e.g. me to spend more time working on 
> Geany). This has to be offset against those that might need to learn 
> e.g. RAII.

Hmmm?..

> * Possibly we might use some STL containers internally.

There things are heavily templated. They generate more code,
compilation is slower, and (unless improved significantly from the last
time I tried them) the code is overall slower than using GArray/GList.
Moreover, you propose to ban "class".

> * Possibly we could use a GObject wrapper for safe reference counting.

Why? Where? How to do that without classes and destructors?
Note that gcc has a non-standard extension for RAII, which works with
the base C types. And Geany compiles with gcc only.

> * Possibly we could have some template function utilities instead of 
> unsafe macros.

Do we need these macros to work with different types? If not, "inline
is already declared in a portable manner in the GLib headers and can be
used normally."

> We would also need to disable some C++ features. I hoped we could do 
> that with g++ flags, but I've only found this so far:
> 
> -fno-rtti (disable runtime info for object inheritance)

+1. Earlier versions of Qt/KDE did that, don't know about current.

> There ought to be a way to disable mixed code and declarations, but 
> maybe not.

Not AFAIK.

> I don't know if this can be enforced by g++, but I suggest we disallow 
> these keywords:
> 
> class

There isn't, that's THE most basic feature of C++.
And you can write classes with struct and union.

> dynamic_cast

-fno-rtti

> friend

No classes -> no usage for friends.

> mutable
> operator

> throw

-fno-exceptions

> virtual

No classes -> no virtual methods.

> Thoughts?

Sorry, but IMHO, that seems even more pointless than using C99.

-- 
E-gards: Jimmy