[Swan-dev] Add -std=c99 to the compiler flags?

[Lennart Sorensen]

On Wed, Feb 04, 2015 at 03:33:50PM -0500, D. Hugh Redelmeier wrote:
> When the code was originally written, there was no c99.  I guess we
> are long past those days and we should start taking advantage of the
> new features.
> 
> I think declaring the variable in the for-head is one of the nicer
> little features.  It goes along with my desire to keep variables as
> local as possible.
> 
> There has been a strong temptation to use the same index variable in
> sibling for statements: it cuts down in the number of declarations and
> lines.  I think this new form eliminates that excuse.
> 
> Why does gcc require a flag?  Is using c99 so remarkable that it cannot be 
> the default?  Is there any downside?  At least some things we already use 
> were introduced in C99.

Well making it the default could make older code no longer compile.
That's pretty bad (and would be rather like what modern C++ compilers
do to older perfectly valid C++ code).

So I think gcc's behaviour for C99 is much better than g++'s treatment
of older C++ code.

> What other C99 features are worth using?
> 
> - local declarations mixed in with statements.  We've avoided this 
>   (mostly) but it is sometimes quite nice.  Actually, this predates C99.
>   This allows us to have fewer variables that are uninitialized.

Certainly something that can have lots of arguments over weather that
is a great idea or terrible code style.

> - I think that we already use the struct and union member initializing 
>   syntax
> 
> - Variable Length Arrays (VLAs) have been made optional-to-support in the 
>   more recent standards so we should not use them.  Too bad: they could
>   do what alloca was used for.

I can't remember what the problem they caused was and why they were
made optional.  Must have been some reason.

> - inline functions -- we already use them.  How come we don't need 
>   -std=c99 already?

Just because they are now part of the standard and hence required to be
supported by a standards compliant compiler does not mean compilers were
not allowed to permit them before hand.

> - <stdbool.h>: we should replace the bool abstraction that I created.
>   Should be trivial
> 
> - <inttypes.h> I think we already use it

Those two are certainly nice things to be able to rely on.

> - // we use this a bit.  I don't like it except in special cases.
> 
> - compound literals: might be useful

OK I am going to have to go look up what that is.  Hmm, having looked them
up I am still not sure what is supposed to be new and exiting about them.

-- 
Len Sorensen