[Info-vax] PL/I (Re: VMS Software needs to port VAX DIBOL to OpenVMS X86 platform)

Tim Sneddon tsneddon at panix.com
Tue Jan 5 02:05:16 EST 2021


Craig A. Berry <craigberry at nospam.mac.com> wrote:
> 
> On 12/26/20 6:59 PM, Rich Alderson wrote:
>> "Craig A. Berry" <craigberry at nospam.mac.com> writes:
>> 
>>> On 12/26/20 9:43 AM, Simon Clubley wrote:
>>>> On 2020-12-26, Craig A. Berry <craigberry at nospam.mac.com> wrote:
>> 
>>>>> On 12/26/20 9:21 AM, Simon Clubley wrote:
>>>>>> On 2020-12-26, Arne Vajh??j <arne at vajhoej.dk> wrote:
>>>>>>> On 12/26/2020 6:16 AM, Simon Clubley wrote:
>> 
>>>>>>>> Is anyone still using PL/I BTW ?
>> 
>>>>>>>> I thought DEC sold off the PL/I compiler and that the person it was
>>>>>>>> sold to retired and either sold off or wound down his business.
>> 
>>>>>>> Kednos & Tim Sneddon?
>> 
>>>>>>> Arne
>> 
>>>>>> Yes, that was him. Thanks for reminding me of the name.
>> 
>>>>> It was Tom Linden, not Tim, who owned Kednos.  I think Tim worked there
>>>>> at some point and kednos.com now redirects to Tim's endlesssoftware.com.
>> 
>>>> Thanks for the correction. It's not a language I have ever used and
>>>> I have long forgotten any detailed knowledge of who PL/I was sold to.
>> 
>>> I've never used it either and I know nothing about current ownership or
>>> plans.  I just remember that it came up in discussions of what gets lost
>>> in a platform port.
>> 
>>> There is a now-moribund PL/I project for the GCC back end:
>> 
>>> <http://pl1gcc.sourceforge.net>
>> 
>>> and there is a proprietary compiler that runs on Linux:
>> 
>>> <http://www.iron-spring.com/about.html>
>> 
>>> Both of these emulate IBM PL/I and would likely require a monumental
>>> effort to replace DEC PL/I.
>> 
>> As I remember the book on the VAX (VMS) port, the goal was very much to
>> "emulate" (implement) IBM PL/I, so I doubt that there would be that much
>> effort for any implementation with the same goal...
> 
> Interesting.  I would have expected a lot of DECisms similar to the
> keywords in BASIC for dealing with RMS files and such.

There are a bunch, but most of them were already there as IBM (the
language inventors) already had ISAM, etc. that needed all those
keywords anyway.

>                                                         But even if the
> language syntax is identical with no extensions, there would still be a
> lot of work to make one of these compilers replace the DEC/Kednos PL/I
> compiler unless the answer to all of the following questions is "no."
> Do you want command-line compatibility?
> Do you want to support the VMS calling standard?
> Do you want to debug the code with the VMS debugger?
> Do you want compatible messages, warnings, and errors?

These are all correct.

> 
> ...and possibly others I'm not thinking of.
> 
> The folks making money from PL/I on OpenVMS Alpha did not see a business
> case for porting it to Itanium almost 20 years ago,

There absolutely was a case.  Unfortunately, there was also an expectation
that somethings should come for free.  That something being the GEM backend.
PL/I uses the BL 21 release of GEM (with some other tweaks from later
versions).  However, there were some big changes in GEM that broke things
so much so that Ada didn't move forward either.  By the time Itanium
happened GEM development had moved on and little thought was given to all
the annoying things that PL/I needed.  So, a PL/I on Itanium either needed
a retargetted compiler or development to bring the VMS code-base to a
later GEM an then a move to GEM on Itanium.

I proposed retargetting the UNIX compiler to LLVM and moving forward.  This
was knocked back because the opinion was that HP should provide the GEM
backend, etc. for free.  There was nothing technical about the decision
at all.

>                                                      so I'm not sure what
> the business case would be now for getting the same or equivalent
> compiler up and running on OpenVMS x86.

Well, a business case is a difficult thing to come up with now.  Unfortunately
those decisions that left PL/I in the dust back in early-mid 2000's meant
that customers got very annoyed.  Many moved or rolled back to the last DEC
release of the compiler.  As smaller customers left the bigger customers
got stiffed with the bill as the run-time license disappeared and customers
were forced to pay for a develpoment license (which could be easily x10-20
what they were already paying).

The only option for Itanium was the binary translator.  Sadly, that was
not a viable option for anyone as the most basic of benchmarks show...

>                                          That said, I'd be delighted to
> see an offbeat and mostly-forgotten language brought back to life by its
> users, if any such folks are still out there.

Me too :-)  Maybe it will happend one day.

Regards, Tim.



More information about the Info-vax mailing list