Re: printf and glibc extensions

[Duane Ellis <duane at duaneellis dot com>]

Eric Blake wrote:
> Is there any interest in the following glibc extensions to printf? 
newlib is nice because it is is specifically _NOT_ the bloated monster 
glibc.

Is the goal a glibc clone? If so, the rhetorical question becomes: Why 
not just use glibc?
And leave newlib alone..

In your defense, standards are always a good thing.

But to answer your question, it depends on your _target_ . 

For embedded, bloat=bad, in my micro controller world: I have 64K of 
flash, nothing more.
Others have megabytes.

Perhaps, a  --enable-newlib-lean-and-mean is needed....

> Obviously, the new extensions would need to be controlled by a configure
> time switch (and perhaps some of the old ones grouped into that switch, so
> that a strict environment has smaller code size).  I'm thinking
> - --enablie-newlib-io-extensions, similar to the existing
> - --enable-newlib-io-pos-args, --enable-newlib-io-long-long, and
> - --enable-newlib-io-long-double.
>   
There are too many --enable-this-funky-feature-options.

Worse, you must dig through levels of ./configure to find the one you need.

And I just suggested '--enable-newlib-lean-and-mean' .....

--Duane.