[Info-vax] HDF5 going going, perhaps soon gone from VMS?

[Arne Vajhøj]

On 8/19/2015 8:09 AM, John E. Malmberg wrote:
> On 8/19/2015 4:21 AM, Dirk Munk wrote:
>> Dirk Munk wrote:
>>
>> Supplying groups like these with free VMS Itaniums is a bit difficult.
>> But couldn't VSI set up an online software development server for
>> projects like these? VSI could set up a complete environment for HDF,
>> set up accounts etc. as a kind of service. They could offer some support
>> for VMS specific questions etc. Projects returning or staying on VMS is
>> always good news I suppose.
>
> HP has(had?) an open source cluster with free access set up for this
> already.  I have not used it for a while because the latency to India
> was too high, and I currently can not reach it.
>
> Free Alpha emulators are also easily available.  And most Alpha source
> will just build and run on Itanium.
>
> The issue is that for VMS support, you need at least one VMS advocate
> dedicating significant time to do testing and development.
>
> You will find that every other OS supported by an open source project
> has at least one maintainer active on the support forum for it.
>
> In the case of Python 3.6x future VMS support being removed that was
> brought up last year, several VMS people on the python indicated that
> they wanted to help.
>
> I set up a cpython repository on vms-ports and did some test builds of
> the development cpython with the VMS support removed.
>
> I had to make some updates to the GNV LD/CC wrapper to get the basic
> build to work, and was able to get the python interpreter building,
> running, and also got most of the dynamically loaded extension libraries
> working.
>
> And most of the stuff that is on my todo list to get things completed
> appears to have never been done for the existing VMS port of Python
> 2.7.x, and are blocking me from using Python 2.7.x for some personal
> projects.  Until that can be fixed, I have to use a Linux host.
>
> None of the VMS people on the Python list have responded to the posts
> that I had a build procedure set up.
>
> The build of Python 3.x only has some very small modifications to the
> existing python code, and those have been submitted as non VMS specific
> patches.
>
> With the updates to GNV, instead of trying to maintain manual edits of
> the upstream library, we are using first include files and supplemental
> libraries to the CRTL as the preferred method, and if a source change is
> needed, a TPU script is used.  Bash needed that.
>
> This allows us to build a current Bash, coreutils, grep, sed, with
> minimal effort when a new release comes out.

If we define 3 groups (overlapping):
* people that want XYZ
* people that have an opinion on how XYZ should be
* people that are willing to spend time working on XYZ
then I think that last group is way too small for VMS.

Arne