Path: utzoo!attcan!lsuc!ecicrl!clewis
From: clewis@ecicrl.UUCP (Chris Lewis)
Newsgroups: comp.lang.perl
Subject: Re: Making PERL@41 on a 3b1 -- taint is dead -- and my dbm.
Message-ID: <973@ecicrl.UUCP>
Date: 30 Nov 90 18:21:17 GMT
References: <1263@kosman.UUCP>
Reply-To: clewis@ecicrl.UUCP (Chris Lewis)
Organization: Elegant Communications Inc., Ottawa, Canada
Lines: 44

In article <1263@kosman.UUCP> kevin@kosman.UUCP (Kevin O'Gorman) writes:
>It looks like taintperl is now thouroughly impossible on a 3b1 -- the
>switch case in teval.c is just too big for the compiler.  I changed it
>to link perl to taintperl.  I've got a single-user system anyway -- why
>should I waste time on security features?

I don't know what your problem is, but we've never had trouble
compiling any patch level of Perl on our 3b1's.  We're 3.5.1.4 release
of the O/S (oldish) with 2Mb, are you something even older?  teval.c
is copied from eval.c and the TAINT definition during compile doesn't
make all that much difference.  Eg: if 320 clauses was too many, you'd
have blown the compile of eval.c too and not been able to compile Perl
at all.

In fact, we use our 3b1's as a yacc server for compiling Perl on
a 386/ix 1.0.6 system because the yacc on that release is too small
to handle Perl.  Mind you, eval.c and one of the other source files
blows the optimizers on a number of O/S's (386/ix 1.0.6 and RS/6000
comes to mind)

And finally, that 320 clause siwtch statement makes me feel good about
psroff - I've had one or two complaints that the 256 clauses in a
switch in psroff is too many ;-)  (it works fine with my 3b1)

>On the other hand, hasn't this gone on long enough?  Larry??

>The results work okay, except that the dbm test fails on 3 -- an attempt
>to do a foreach on an empty array.  I'm not really sure what I'm using
>for dbm... Anyone know where there's a good one to snarf?

What did it use?  Chances are it didn't use any (eg: all assoc arrays
incore) unless you explicitly imported mdbm or ndbm because ol' dbm
doesn't come standard on 3b1 releases that I know of.  On that front,
it appears that Ozan's PD sdbm will be a freely redistributable full-feature
dbm that works with Perl (don't ask him or me for it because it's still
under test and needs a little work before release)

BTW: I'm getting core-dumps with PL41 too (though the regression tests
work perfectly), am still investigating.
-- 
Chris Lewis, Phone: (613) 832-0541
UUCP: uunet!utai!lsuc!ecicrl!clewis
Moderator of the Ferret Mailing List (ferret-request@eci386)
Psroff mailing list (psroff-request@eci386)


Brought to you by Super Global Mega Corp .com