any plans for an "official" newlib snapshot ?
Mon Nov 20 14:11:00 GMT 2000
J. Johnston wrote:
> Yes there is. I would like to incorporate your linux work into the
> base as well as the long double printf/scanf solution before creating the
> next snapshot.
Okay, the status of my Linux patch is that basic things work, but some
things are missing, and other things are untested and may be buggy.
For missing things, what are the general newlib goals with respect to
ANSI C libc or POSIX 1003.1 complicance ? Is the short-term goal
basically to have a "reasonable" subset or are there more specific
Concerning untested things, I'd really have to start with a regression
test to get good coverage here. Also, header files probably don't look
too good when it comes to name space pollution. Generally, I currently
fix problems as I hit them, but I don't search for them.
My work is mainly driven by the requirements of an initrd environment.
The next steps are:
- make it work nicely with bootimg (a very Linux-specific function)
- add networking (goal: be able to mount NFS file systems)
- do whatever it takes to get programs using autoconf to build (within
limits, of course)
I think none of this is of immediate interest for newlib, so it might
be best to merge the current Linux code as a baseline (that is, after
fixing the FIXMEs), and flag it as "experimental", while I'll continue
along the lines above.
Does this sound okay ?
/ Werner Almesberger, ICA, EPFL, CH Werner.Almesberger@epfl.ch /
More information about the Newlib