Remaining tasks for 1.1.4 release as of Jul 26

Remaining tasks for 1.1.4 release as of Jul 26

[Rich Felker]

Here are the only major things I see left for the 1.1.4 release:

- Adding non-dummy gettext functions
- Finish reviewing and merging if_nameindex/getaddrinfo patch
- Fixing misplaced/missing barriers for SuperH atomic asm
- Merging some remaining atomic asm fixups I have in progress
- Reviewing and merging getopt_long abbreviated options patch

I had IDN support on the Roadmap agenda for this release cycle but I
just pushed it back because all of the other items are basically done
and I don't want to push the release back to start on something
completely new. So it's on the 1.1.5 agenda now. There's also Felix's
request for getopt_long permutation that I may do, but I think it will
need some time to study and implement so let's look at it after
release too. Likewise, the there are two other remaining patches from
Alpine that are probably close to ready for upstream but that I'd like
to deal with after release.

Anything else I'm missing in the way of important open issues that
haven't been resolved?

Rich

Re: Remaining tasks for 1.1.4 release as of Jul 26

[Rich Felker]

On Sat, Jul 26, 2014 at 04:14:48PM -0400, Rich Felker wrote:
> Here are the only major things I see left for the 1.1.4 release:
> 
> - Adding non-dummy gettext functions

This is "done", but there are some additional features I'd like to get
in before the release: $LANGUAGE support and perhaps plurals. nsz
mentioned on IRC possibly writing an expression evaluator for this but
I don't know the timeframe.

> - Finish reviewing and merging if_nameindex/getaddrinfo patch

I've sent my remaining comments on this to the list. I'd like it if we
could get this smaller/simpler but I don't see any good way.

> - Fixing misplaced/missing barriers for SuperH atomic asm

Done. This code still has a number of issues I'd like to clean up, but
at least it should be safe now. In the long term I'd like to use a
similar strategy for both ARM and sh (and any other archs where the
atomic implementation needs to be selected at runtime) to minimize
both the code size/register spillage in the caller and the run time in
the callee.

> - Merging some remaining atomic asm fixups I have in progress

I've done the cleanup of cruft. I want to change some constraints to
be more consistent and add volatile attributes, but it's enough work
and enough chance to mess things up that I think I'll punt on the rest
for now and revisit it later.

> - Reviewing and merging getopt_long abbreviated options patch

I'm looking into a simpler approach to this that avoids the problem of
matching ambiguous abbreviations.

Rich