Re: VMS-Perl and VSI-Java

Craig Berry via vmsperl <[email protected]> Thu, 24 Oct 2019 21:26:58 -0500
Newsgroups gmane.comp.lang.perl.ports.vms
Message-ID <[email protected]>
> On Oct 22, 2019, at 7:30 AM, Craig Berry via vmsperl =
<[email protected]> wrote:
>=20
>> On Oct 21, 2019, at 2:49 AM, Jouk Jansen <[email protected]> =
wrote:
>>=20
>> I noticed problems with Perl, when JAVA80 is configured. I tracked =
down the
>> problem to some decc$xxxx logicals, which are defined by the java =
setup.
>=20
> Yes, I ran into this when the first HPE/VSI Java kit came out.=20
>=20
>> Maybe it is a good idea to define the required logicals inside the =
perl.exe.
>> An example how to do this can be found (i.e.) in the sources of bzip2
>> (http://antinode.info/dec/sw/bzip2.html) (see file vms_misc.c)
>>=20
>> What do you think of this idea?
>=20
> Perl already has a lib$initialize set-up and does enable the things =
that it needs:
>=20
> <https://github.com/Perl/perl5/blob/blead/vms/vms.c#L13835>
>=20
> (Yes, we moved to GitHub recently).
>=20
> I guess we could also make Perl disable things that are enabled by the =
Java command procedures and break Perl.  This seems backwards to me and =
it is really Java that should enable what it needs at run time rather =
than requiring anyone who has Java installed to take evasive action.  If =
I remember correctly, Java set-up actually does DEFINE/SYSTEM so =
ordinary users can't even deassign the problem logicals but have to =
override by assigning different values.
>=20
> Thanks for the suggestion.  I'd be happy to hear what others think.  I =
was hoping HPE/VSI would fix Java, but at this point there will likely =
never be another truly new Java kit on Itanium, even if there is a minor =
update to the current one.

By the way, the only one I've run into trouble with is =
DECC$READDIR_DROPDOTNOTYPE.  Are there athers?  And it looks like I =
remembered wrong, and there are ways to have Java put the logicals in =
the process table rather than the system table, but still it seem it =
should just set the features for the duration of the Java image and not =
pollute logicals that could affect anything else.

________________________________________
Craig A. Berry

"... getting out of a sonnet is much more
 difficult than getting in."
                 Brad Leithauser