That's funny, because earlier someone said that it did.
Sent by Magic!
> On Jul 10, 2016, at 12:42 PM, Rick McGuire wrote:
>
>
> ooRexx already does not allow Rexx queue communication between different
> uids.
>
> Rick
---
On Sun, Jul 10, 2016 at 2:03 PM, CVBruce wrote:
> I still have concerns about this, and please don't take this as vote
> against, it isn't. In my last email I specifically mentioned the Windows
> implementation. I still haven't seen any mention on how we are going to
> make Windows work the sam
I still have concerns about this, and please don't take this as vote against,
it isn't. In my last email I specifically mentioned the Windows
implementation. I still haven't seen any mention on how we are going to make
Windows work the same way as *NIX.
We all know the saying Rexx Anywhere,
This looks interesting. I think it would be better to refactor the code so
that the domain sockets are implemented by a separate class, leaving the
SysCSSStream class still available if we have platforms where domain
sockets are not an option. This was a much smaller change than I expected.
Rick
Currently, the only difference between the API daemons is how they are
started. On Unix, they are designed to be started as a system service that
opens a TCP port for all ooRexx clients to connect to. As a TCP port can
only be opened once the daemon is shared among all users. I think on
Windows, Ma
I guess I want to understand more of the details. What is the Windows
implementation? What are the differences in usage, abilities, and constraints
with the current implementation.
Sent by Magic!
> On Jul 9, 2016, at 12:46 PM, Moritz Hoffmann wrote:
>
> Going forward, I implemented unix doma
Going forward, I implemented unix domain sockets as a means of
communication on Linux. See attached patch. I didn't commit it as we first
need to reach consensus whether it should go into ooRexx 5 or not, but I
think it should. It would be great if someone else could adapt the Mac and
AIX code to m
De: Moritz Hoffmann
I think a clean solution would be to use UNIX sockets on Linux and similar
systems. I don't know about Mac or Windows but I'm sure there exists something
equivalent. Then, each user would have its own instance of the api daemon that
is accessible using a socket stored in th
Greetings Rick,
Rick McGuire wrote:
> The pre-4.0 rxapi used named shared memory that had a lot of inherent
> stability problems. Because the shared memory used a unique name, it also
> suffered from the same problem. I would definitely NOT
> recommend returning to that mode of inter-process com
On Wed, Jul 6, 2016 at 2:20 PM, Moritz Hoffmann wrote:
> As far as I remember, there was no api daemon back then. Instead, ooRexx
> used shared memory to enable communication between different instanced
> belonging to the same user. This is definitely not the way we want things
> to work these da
As far as I remember, there was no api daemon back then. Instead, ooRexx
used shared memory to enable communication between different instanced
belonging to the same user. This is definitely not the way we want things
to work these days as it is hard to maintain and prone to bugs. The current
desig
On Wed, Jul 6, 2016 at 1:47 PM, Michael Lueck
wrote:
> Greetings all,
>
> I was thinking to ask to compare to the pre-ooRexx v4 way of handling
> RXAPI, however...
>
> Moritz Hoffmann wrote:
> > I think a clean solution would be to use UNIX sockets on Linux and
> similar systems. I don't know abo
Greetings all,
I was thinking to ask to compare to the pre-ooRexx v4 way of handling RXAPI,
however...
Moritz Hoffmann wrote:
> I think a clean solution would be to use UNIX sockets on Linux and similar
> systems. I don't know about Mac or Windows but I'm sure there exists
> something equivale
There's one problem that we're currently not looking at: The api daemon
uses the user name as identifier to separate between different users. The
name is sent in a message from the client to the server which means that it
could be forged, i.e. any user can access any other user's data. It is not
po
Yes, both ooRexx client program and rxapi would have to know where to look for
the PID file. Unless there is a pre-arranged fall back place, something like
~/.ooRexx/rxapi.pid
> On Jul 6, 2016, at 3:25 AM, Rick McGuire wrote:
>
>
>
> On Wed, Jul 6, 2016 at 6:18 AM, Rony G. Flatscher
That could work, but I don’t know that a second file would be needed, couldn’t
the PID file contain both PID and port? I’m also assuming that you would
determine PID by matching effective user ID’s of both ooRexx client and rxapi.
Also going forward this would be a solution for V5, I don’t see
On Wed, Jul 6, 2016 at 6:18 AM, Rony G. Flatscher
wrote:
>
> On 05.07.2016 22:43, CVBruce wrote:
>
> Although this gets around the pid file problem, it doesn’t get around the
> port problem, you can still only run one rxapi process per computer. For
> example you can’t run both 32bit and 64bit
On 05.07.2016 22:43, CVBruce wrote:
> Although this gets around the pid file problem, it doesn’t get around the
> port problem, you can still only run one rxapi process per computer. For
> example you can’t run both 32bit and 64bit rxapi concurrently on the same
> box, even if your have differ
18 matches
Mail list logo