A todo update from larry to acknowledge the existance of hush

This commit is contained in:
Eric Andersen 2001-05-03 17:35:48 +00:00
parent 87d893c04b
commit b0aca921be

23
TODO
View File

@ -10,15 +10,13 @@ around to it some time. If you have any good ideas, please let me know.
netkit-tiny. Per discussions on the mailing list, this isn't going
to happen. False alarm. Sorry about the confusion.
* The busybox shell, lash, is really too weak for serious use, although it is
possible to run many simple systems with it. BusyBox 0.49 was supposed to
have a new shell, updated to understand full Bourne grammar. Well, that
simply didn't happen in time for the release. A rewrite is in progress
that will result in a new shell that understands the full Bourne grammar.
This new shell is being championed by Larry Doolittle
<ldoolitt@recycle.lbl.gov>, and could use your help. Please see the work
in progress at http://doolittle.faludi.com/~larry/parser.html and help
us out.
* The default busybox shell, lash, is really too weak for serious use,
although it is possible to run many simple systems with it. BusyBox
0.52 now contains a rewritten shell, hush, which understands most
Bourne grammar, with only about a 4 Kbyte binary size penalty. You can
engage hush at pre-compile time by "ln -sf hush.c sh.c; touch hush.c".
Hush is young, and has plenty of bugs to shake out, so think twice before
using it for production systems. We welcome bug reports and patches.
-Erik
@ -31,8 +29,11 @@ Possible apps to include some time:
-----------
Write a fixup_globals function to do just that right before calling
non-forking applets. Or, just always fork in the shell...
With sysvinit, reboot, poweroff and halt all used a named pipe,
/dev/initctl, to communicate with the init process. Busybox
currently uses signals to communicate with init. This makes
busybox incompatible with sysvinit. We should probably use
a named pipe as well so we can be compatible.
-----------------------