Snapshot vs. CVS
Wed Sep 28 21:49:00 GMT 2005
Jonathan S. Shapiro wrote:
> I'ld like some advice -- most likely from Jeff but other input is
> When I pull in a subsystem for integration, my usual practice is to try
> to stick to "releases" and patch minimally. For newlib this means that I
> picked up the 1.13.0 tarball. So far, ignoring my port-specific patches,
> the only patch I have needed has been the "_ctype_b" patch.
> I've just run into the stdint.h issue. I need it, and looking over the
> mailing list I can see that (a) this is a current issue under
> discussion, and (b) it's addressed in the CVS tree.
> I'm tempted to move forward to the CVS current branch, but I don't have
> time to be patching this every day. Which raises two questions:
> (1) How stable is the CVS tree at the moment? I.e. how often do I need
> to be planning to update?
Let me start by addressing the stability of the tarballs. The tarballs
are simply snapshots of the tree usually once a year (December). The
tarball is also usually tested by the RTEMS folks and if they raise a
flag, I will rebuild it. There is no special stress testing on this end
other than I will ensure the x86 linux builds and can run some sample
The CVS tree is being constantly used by the Cygwin folks in their
nightly build and they usually speak out immediately when something
causes a problem in the shared code. We don't have experimental phases
like you might find in gcc or other projects (e.g. a new parser being
tried out). It is a little more straightforward because we are for the
most part dealing with functions that are well-defined.
If the December time frame works for you, then start using CVS knowing
that 1.14 will be a snapshot of CVS in December.
> (2) Do you have an approximate sense of when the next snap is likely to
> be taken? I suppose I'm really asking: are things presently in an
> "active development" phase, or are they anticipated to be entering a
> "trying to coverge to snapshot" phase some time soon?
December time-frame for 1.14. Things presently are in normal state
whereby any fixes / enhancements are looked at. I usually announce the
snapshot is coming and those who have been sitting on fixes are
encouraged to send them in. In the end, I don't discourage any fixes to
be submitted when I'm about to cut the snapshot but eventually I make
the cutoff. December is usually slower anyway because of the upcoming
-- Jeff J.
> I can handle it either way. I'm trying to assess what sort of time
> allocation I need to assign to newlib.
More information about the Newlib