[Info-vax] Python for OpenVMS VAX

[Arne Vajhøj]

Richard B. Gilbert wrote:
> Chris Allen wrote:
>> On Oct 13, 3:23 pm, sapienzaf <sapie... at noesys.com> wrote:
>>> On Oct 13, 5:43 pm, Chris Allen <ca.al... at gmail.com> wrote:
>>>
>>>> Unfortunantly I can't move to Alpha or Itanium,
>>>> otherwise I would have already.
>>> May I ask why not?  Is it a cost issue, management inertia, or
>>> something else?
>>>
>>> If you have the source code then I would think porting becomes a very
>>> good option.
>>
>> There's a few reasons.  We run quite a bit of legacy software that was
>> written by an employee who is deceased.  It might be a snap to port
>> his software to Alpha or IA64 but we don't know.  His programs run
>> great, and haven't needed any modifications in years.  Also we run a
>> very ancient source code control tool called CCC (this predates CCC
>> Harvest), for which we only have a VAX binary of.  My goal as a system
>> manager here is to get off of CCC entirely, so not being able to run
>> CCC on the new machine won't be a problem if I can migrate the code
>> out of it into say Mercurial.  Now that I'm thinking about it it might
>> not be a big deal to move to Alpha or IA64.
> 
> VMS to Alpha is generally not a difficult port.  About the only 
> potential problem that I can recall is a program that has a dependency 
> on the size of a "page" of memory.  I don't recall the details but VAX 
> pages are 512 bytes and Alpha pages are bigger; I don't recall how 
> *much* bigger.

8192

(the architecture supported bigger, but I believe that is what all
Alpha systems in existence used)

>               Alright, one more potential problem!  If you are doing 
> bit twiddling in floating point variables, the format differs between 
> VAX and Alpha.

Only if using IEEE floating point.

F and G VAX floating point behave like they always have.

(D VAX floating point is a bit messy)

> Most programs are compile, link, and go.

Yep.

Arne