develooper Front page | perl.perl5.porters | Postings from November 2000

Re: [PATCH: perl@7674] updates to README.os390

Thread Previous | Thread Next
Peter Prymmer
November 14, 2000 11:49
Re: [PATCH: perl@7674] updates to README.os390
Message ID:

On Tue, 14 Nov 2000, Prymmer/Kahn wrote:

> it does at least update the current README.os390 file a bit (although
> some serious problems remain in the C<use utf8;> department for the port
> at this time).

On Wed, 1 Nov 2000 Jarkko Hietaniemi had suggested:

! Why not
! #ifdef EBCDIC
! dowarn = 0;
! #endif
! or, actually
! #ifdef EBCDIC
! flags = UTF8_ALLOW_ANY;
! #endif
! right in the beginning of the utf8_to_uv()?

and the answers to each of these is as follows: with the latter
C<flags = UTF8_ALLOW_ANY;> the compilation of utf8.c fails since
that statement seems to tickle what c89 thinks is a syntax error (I'd post
the message but I don't have that build set up right now).

With the former C<dowarn = 0;> the compilation proceeds fine through
everything, but `make test` hangs at one of the first tests that seems to
exercise the C<use utf8;> pragma to any appreciable extent, I obtain:

 % make test > make.out 2>&1 &
 ... time passes ...
 % tail make.out
 comp/proto...........FAILED at test 111

and in this case the time that passes was on the order of 18 hours.

I look forward to testing the patch that Yitzchak Scott-Thoennes sent in

the message "Malformed UTF-8 character" was what we had been seeing
in comp/require.t and several other tests on OS/390 recently.  In fact, 
if I let utf8.c alone and do no patching to the 7646 kit and run:

   % make test > make.out_0 2>&1

then the process does complete (does not hang) but make.out_0 winds up
being quite large and full of spurious warnings:

   % -rw-r--r--   1 PVHP     OMVSGRP  117849921 Nov 13 14:11 make.out_0
   % grep Malformed make.out_0 | wc -l

Peter Prymmer

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