Hi,
After trying to fix a bug in uClibc-ng, which always let tst-kill4
segfault on my system, I rethought my spin-off. I made a lot of
mistakes in my spin-off. First I didn't include the git history
from uClibc, so I cannot use git bisect, git log and git blame to
find more information about any changes made from uClibc to
uClibc-ng. Second I started to run embedded-test.sh after adding
and removing functionality in uClibc-ng. Git master of uClibc is
worse than I thought of. I think nobody regulary runs any compile
and runtime testing on the supported architectures.
Furthermore I've get a little bit sentimental after getting my hands
on a Cisco 1000 series router, which after nearly 15 years still
boots up fine. The soul of embedded systems exist, so I shouldn't
remove support for embedded vintage hardware in uClibc-ng ;-)
Instead I will try to fix open issues and try to run uClibc-ng on my
old embedded systems including Foxboard LX (cris) and NGW100
(avr32).
I will soon have more space for my embedded system collection in a
new hobby room.
So my plan for uClibc-ng is:
1. Restart the spin-off from uClibc master and use git to manage it
in a better way (done with a lot of help by Thorsten Glaser)
2. Fix open issues in uClibc and report them to the uClibc project
(started with a sparc pipe() fix recently)
3. Run the uClibc test suite and pubish the results for all architectures
I either have an emulator or hardware
4. Deactivate some architectures, which can not be tested for 1.0
5. Start to add bugfixes from the different projects (OpenWrt, buildroot, openembedded,
...)
6. Add following features: $ORIGIN support, libuargp and Xtensa NPTL
7. make a release
have fun
Waldemar
Show replies by date