[Info-vax] VMS and package managers, was: Re: Micro Focus Merger with Hewlett Packard Enterprise's Software Business Segment

[John E. Malmberg]

On 10/4/2016 7:40 AM, Simon Clubley wrote:
> On 2016-10-04, John E. Malmberg <wb8tyw at qsl.net_work> wrote:
>> On 10/3/2016 7:57 PM, Simon Clubley wrote:
>>
>>> If VMS had some kind of decent init system for startup, such a
>>> package manager could then automatically insert the package into
>>> the startup sequence (and at the correct point) if required and
>>> then remove it automatically during product removal.
>>
>> I do not know what your definition of decent is.  The sysman utility can
>> update that startup database:
>>
>>    mcr sysman help startup
>
> Decent means being able to enforce relationships between packages
> by means of startup sequencing; sysman has absolutely no concept
> of inter-package sequencing as far as I can see.
>
> For example, the HP C compiler requires a manual change to systartup_vms
> to start HP C _after_ startnet.com as shown here:
>
> http://h41379.www4.hpe.com/commercial/c/docs/ig_bvbv_001.html#systartup

That appears to be a documentation bug.

There is nothing in the decc$startup.com that cares if it is before or 
after the network startup.  All it is doing is installing two images.

I figured that was a bug before I even looked at the decc$startup file.

A more complex package would be one that contains a more complex product 
similar to a CIFS server.

1. The images and data files need to be set up before network service 
component is started.

2. The network service component can not started until after the 
networking is started.  And probably should wait until some dependent 
network services have started.

> and requires the system manager to manually edit the startup file,
> find the right place to manually add the required command, while
> hoping the system manager hasn't either made a typo or forgotten to
> test the changes (people _do_ test their startup changes, right ? :-)).

My sy*.com files can all be re-run on a running system for changes that 
do not need a reboot to take effect.

I also follow a structure similar to the DecStart training that used to 
be sold with VMS installs.  Unfortunately the systartup_vms did not 
follow that model.  It makes the system startup files much more manageable.

> A decent init system handles all that for you and neither sysman
> (as far as I can see) or systartup_vms.com falls into that category.

While Linux appears to have an init system, it is currently being 
tweaked significantly with each distro at a rate where the online 
instructions on how to implement things can be significantly different.

I am currently using at least 5 different Linux distros or release 
versions with a combination of vendor supplied packages and locally 
built packages.

The location of where the "init" files are located for a package vary 
greatly as many times the Linux distros place them in entirely different 
locations than the upstream package.

You edit things the way the upstream says to, and that file is ignored.
You do a search on the issue, and you find pages of other people asking 
the same question with no answer.

Getting some packages to be stable can be fun when the default package 
install for that distro is less than optimal.

Regards,
-John
wb8tyw at qwl.net_work