Rick,
thank you very much for your answers and insights!
---rony
On 05.07.2016 23:04, Rick McGuire wrote:
>
>
> On Sat, Jul 2, 2016 at 11:44 AM, Rony G. Flatscher <rony.flatsc...@wu.ac.at
> <mailto:rony.flatsc...@wu.ac.at>> wrote:
>
> Changed the subject to reflect this thread of discussion.
>
> The desire of many: allow ooRexx to be run off an USB stick to show
> off/demonstrate ooRexx to
> others and/or have a great tool on ones hands to bring along when
> checking out problematic
> computers or helping collegues and friends.
>
> Currently this is not possible on computers, where the user has no
> superuser power over the
> computer, which is a great pity. This scenario is quite usual, when
> computers are deployed in
> business environments, where root/superuser power is reserved for the
> installation,
> maintenance teams and the like.
>
> The reason for this restriction seems to be how rxapi gets started and
> used.
>
> So maybe a few questions therefore about rxapi:
>
> * What is the rationale of allowing only one rxapi running per computer?
>
> This is largely a throw back to how the original OS/2 version, but there are
> other limitations
> that could make it difficult design (see below).
>
> *
>
>
> * What is the rationale to allow to install it with root/superuser
> rights only?
>
> David Ashley did all of that work, but my understanding is it was driven by
> linux conventions for
> daemons that are launched at system boot. The rxapi server is quite capable
> of being launched on
> demand, but it is still limited to one instance per system.
>
> *
>
>
> * Would it be theoretically possible to run separate instances of rxapi
> for different user
> sessions (one per session) by default, one may be a
> root/superuser/daemon session?
>
> The current design uses a classic tcp client/server architecture for handling
> requests. The server
> listens on a known port and each new process where Rexx is running
> establishes a connection to the
> server using the defined port. The server spins of a thread to service each
> connection and all
> messages between the client and server are sent on that connection. I don't
> know of any way using
> tcp for a different listener port to be used for each user and for the
> clients run by that user to
> discover the port used by its server.
>
> There may be other possible ipc mechanisms that could be used, but I'm not
> aware of them.
>
> Rick
>
>
> *
>
>
> Ad port number: even though there is a reserved port for ooRexx, this
> does not guarantee that
> it would be available at all on any computer. So in the case of a port
> clash (port already in
> use by a different program at rxapi startup) communicating the rxapi port
> to Rexx clients
> needs to be done one way or the other.
>
> Ad users killing rxapi and thereby affecting running ooRexx programs in
> multiuser
> environments: this would be possible for sudo users at all times and
> doing it is even
> mandatory for upgrading ooRexx currently. (In a normal use case of ooRexx
> programs, users
> would not even know about rxapi so would not know to kill it.)
>
> ---rony
>
> P.S.: Even if this is not tackled for 5.0, it would help deciding changes
> for the future, if
> more information was available about the rationales behind rxapi. It for
> sure would help
> spreading the word about ooRexx by demonstration of its capabilities
> tremendeously.
>
>
> On 01.07.2016 16:57, René Jansen wrote:
>> Hi Erich,
>>
>> I’ll work on trying to get these packages available, at least the ones
>> that Erico does not
>> provide.
>>
>> Yes, we should not change anything for this release. I entirely agree.
>> Still, I am not
>> convinced that ooRexx needs a designated port at all; much less a deamon
>> process running as
>> root. But we will study and discuss in due time.
>>
>> I’ll try to fix the Z installation. Did not know about the rexx.cat
>> <http://rexx.cat> bug.
>> Indeed, file rexx.cat <http://rexx.cat> now reports:
>> rexx.cat <http://rexx.cat>: Nazgul style compiled message catalog,
>> version 1
>>
>> instead as data what it used to do.
>>
>> best regards,
>>
>> René.
>>
>>
>>> On 1 jul. 2016, at 16:45, Erich Steinböck <erich.steinbo...@gmail.com
>>> <mailto:erich.steinbo...@gmail.com>> wrote:
>>>
>>> afaik that is done with cpack and documented in Cmake-build-readme.txt
>>> Thanks for the pointer. These package files are exactly what we should
>>> make available to
>>> our users for beta-testinI do not know the cross-process requirements
>>> of rxapi
>>> rxapi listens on port 10010. It is a system-wide demon - there can
>>> only be one demon
>>> listening on port 10010. 10010 is the official port for ooRexx,
>>> see
>>> https://en.m.wikipedia.org/wiki/List_of_TCP_and_UDP_port_numbers#Well-known_ports
>>>
>>> how could we localize that, possibly running over different port
>>> numbers or just
>>> sockets/pipes localized to the user
>>> I cannot imagine anyone doing a change of this magnitude - we really
>>> should push for beta
>>> and then release
>>>
>>> I also looked at the Z build - could you tell me why this is related to
>>> running from the
>>> build dir?
>>> the test is running off build/bin, but the rexx.cat <http://rexx.cat>
>>> it sees, is from an
>>> older "make install" version. We had a bug concerning how rexx.cat
>>> <http://rexx.cat> was
>>> being built, that I just recently fixed - that's why the two versions
>>> are different and
>>> that's why a few tests fail, which check for exact message texts
>>> There may be other things that are being picked up from an installed
>>> rexx when running from
>>> build/bin, but I haven't checked that
>>>
>>> Erich
>
------------------------------------------------------------------------------
Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San
Francisco, CA to explore cutting-edge tech and listen to tech luminaries
present their vision of the future. This family event has something for
everyone, including kids. Get more information and register today.
http://sdm.link/attshape
_______________________________________________
Oorexx-devel mailing list
Oorexx-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oorexx-devel
