[BusyBox] Segfault in sed during binutils build.
Rob Landley
rob at landley.net
Mon Sep 15 01:17:21 UTC 2003
On Sunday 14 September 2003 20:25, Glenn McGrath wrote:
> On Sun, 14 Sep 2003 19:51:08 -0400
>
> Rob Landley <rob at landley.net> wrote:
> > So, the next question is: are you happy tackling the rest of the cruft
> > in the original testcase I posted (which is an actual live
> > in-the-field sed command being done by the binutils build), or should
> > I bang on it some more?
>
> Im happy fixing defined bugs with nice testcases, but its taking me a
> few hours to isolate each bug, pretty frustrating.
>
> I did start adding some new tests to the testsuite after starting to
> fixing one bug be creating another.
>
> I think the logical thing to do with sed is to spend time adding more
> tests rather than tracking down bugs in the wild.
>
> But adding tests isnt much fun either, and sed would require quite a few
> tests cases to cover it.
>
>
>
> Glenn
Well, my goal here is to build a working linux from scratch distribution with
uclibc and busybox in place of as many packages as they can replace. (And
minus a few other things I don't want, like internationalization.)
I actually want to make a development system using busybox, which means that
I'm trying to compile packages with busybox. And this runs into autoconf,
which seems to lean pretty heavily on awk, sed, and grep...
On the bright side, sed is fairly well documented as to how it SHOULD work,
and I've got gnu sed to produce the output the package is expecting for
comparison purposes, and the sed source is only about 1000 lines (of which
I've now read about 2/3 pretty closely) so... :)
I just don't want to spend my time reinventing the wheel. But if you're not
interested in going down this particular rathole... :)
Rob
More information about the busybox
mailing list