Irek Szczesniak <iszczesn...@gmail.com> wrote:
> Well, I'm strongly against importing more tools from FreeBSD. the
> damage caused by Garrett D'Amore eager "adopt FreeBSD tools to fill
> our needs" project has created gaping holes in
> functionality/interoperability and as side effect imported all the
> damn FreeBSD bugs, too. IMO the changes to tools like tr, sed and many
> others must be reversed because multibyte character handling is
> grossly broken now and all the other issues caused by that.
I am not convinced that this is a result of importing things from FreeBSD.
The tools in question did e.g. pass a POSIX compliance test on Mac OS X...
There is however some kind of common thread with regard to software quality if
we look at work results from Garrett D'Amore:
- After Garret replaced the closed source i18n code in libc by a self
hacked version from FreeBSD, "col -x" no longer passes japanese
characters. This (illumos specific miss-behavior of "col") is why they
are currently discussing whether they should replace the rock solid
nroff based man subsystem by a new and not sufficiently tested
software from OpenBSD.
- The "od" OSS replacement written by Garret is full of bugs.
I would guess that if there are problems with tr and sed on Illumos, this are
problems caused by the i18n implementation they are based ob because they are
linked against the libc from Illumos rather than problems caused by the tools
themselves.
So it seems that it would be sufficient to introduce a working code review for
Illumos. Currently, a failed code review (as happened for Garret's "od") does
not prvent that piece of code from being integrated into the Illumos code base
and it seems that for the i18n replacement code no code review happened at all.
It seems to be also important to forbid code changes as a result of a so called
bug in some other software before the bug was fixed. I am convinced that it
would be sufficient to fix the behavior of the i18n implementation on Illumos
to get rid of problem in other software that is based on the correct behavior
of the i18n functions in libc.
Jörg
--
EMail:jo...@schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
j...@cs.tu-berlin.de (uni)
joerg.schill...@fokus.fraunhofer.de (work) Blog:
http://schily.blogspot.com/
URL: http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily
_______________________________________________
belenix-dev mailing list
belenix-dev@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/belenix-dev
