develooper Front page | perl.perl5.porters | Postings from October 2001

Re: [PATCH hints/] Examining Redhat patches

Thread Previous | Thread Next
Jarkko Hietaniemi
October 30, 2001 07:38
Re: [PATCH hints/] Examining Redhat patches
Message ID:
On Fri, Oct 26, 2001 at 09:42:12PM -0400, Michael G Schwern wrote:
> Out of curiousity, I pulled the Redhat 5.6.0-12 source RPM and looked
> at what they've patched.  Some of it is very RH specific, but there's
> a few little corrections we could include.

I now reviewed them a little bit more .

> The include Digest::MD5, we include Digest::MD5.  Good.


> perl-5.6.0-errno.patch tears out get_files() and process_file() and
> replaces it with a very RH specific get_errors() which goes through
> the defines in errno.h.  I'm curious why the original functions didn't
> work.  Patch attached.

I think they just wanted to use the gcc -dM, which 5.6.0 didnt do,
but we do now.

> >From perl-5.6.0-fhs.patch below is a small correction to
> hints/ that RH uses -devel not -dev.

I still need some with RH history books to find out when
did this change.

> perl-5.6.0-installman.patch appears to be a real problem in pretty
> much everything before 5.7 (ie. INSTALLMAN*DIR not honoring PREFIX)
> but seems finally fixed in bleadperl.


> perl-5.6.0-nodb.patch seems to be thrashing around making NDBM_File
> working with... something.  I've attached it if someone wants to
> figure out if it's useful.

I think the current scheme in ext/NDBM_File/hints/ says to use
gdbm to emulate ndbm if available, and I would think gdbm is always
available in RH, so what they are doing with db, we shouldn't try to

> perl-5.6.0-prereq.patch is a very coarse hack to have Makemaker dump
> out the module prereqs and then exit.  Something like this might be
> useful as perhaps a make target or MakeMaker function.  Attached.

This still looks reasonable.  Anybody feeling like looking into this?
(The RH patch might be enough on its own)

> perl-5.6.0-syslog.patch looks like it's adding syslog logging to
> validate_suid().  It looks like a lot of the code it's emulating (like
> sending mail to root) is gone.  I've attached it for someone more
> knowledgable about suid to evaluate.

Not touching it given the steady supply of syslog() overflow
incidents.  If I touch the suid code anymore, it's hopefully
only to dissect it away.

> The rest is all RH stuff.

Could someone with access to a RH box (or preferably, more than one
RH), try out the latest blead, with an eye towards the areas the above
changes tried to address?

$jhi++; #
        # There is this special biologist word we use for 'stable'.
        # It is 'dead'. -- Jack Cohen

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About