Re: Calling a remote kernel from an external program.
- To: mathgroup at smc.vnet.net
- Subject: [mg27891] Re: Calling a remote kernel from an external program.
- From: Jens-Peer Kuska <kuska at informatik.uni-leipzig.de>
- Date: Fri, 23 Mar 2001 04:31:10 -0500 (EST)
- Organization: Universitaet Leipzig
- References: <98vik0$938@smc.vnet.net> <9962nq$t2i@smc.vnet.net> <99chdc$8bd@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Hi, I tryed on my SGI box and all works fine Session 1--------------------------- lnk = LinkCreate["8000"] Session 2--------------------------- lnk = LinkConnect["8000@phong"] LinkWrite[lnk, 15!] Session 1--------------------------- LinkReadyQ[lnk] LinkRead[lnk] 1307674368000 The opposite version, create the link in Session 2 and connect to the link in session 1 works as well. What Operating System you are using ? You should try some other protocols for your system. Perplex Jens parra at signition.com wrote: > > In article <9962nq$t2i at smc.vnet.net>, Jens-Peer Kuska says... > > > >Hi, > > > >if you can call LinkCreate[] than you have a program that > >acts like a FrontEnd and the kernel can't listen to > >two fontends you must call the kernel with > > > >math -mahtlink -linkcreate > > > >or reset the parent link to your program > > > >Regards > > Jens > > > > > >PS: > >Ok. The kernel can listen to two fontends with the > >brilliant trick invented by Tod Galley in it's J/Link > >packages. > > > > Thanks for the reply, > > I forgot to mention that I reset the parent link to the result of > LinkCreate and still no go. So I tried a simple example of running > two sessions (from 2.12.6) on the same machine and try to have them talk. > On one session I run > > In[1]:= link = LinkCreate["8000"] > > Out[1]= LinkObject[8000@nambe, 1, 1] > > On the other session I do > > In[1]:= link = LinkConnect["8000"] > > Out[1]= LinkObject[8000@nambe, 1, 1] > > Then on the first session I do > > In[2]:= LinkWrite[link, 15!] > > and on the second > > In[2]:= LinkRead[link] > > At this point nothing happens. The second session never returns the expected > value. I tried other port numbers but nothing works. If I control-C on session > one, both seesions die with a message saying something about a pollable event, > so the two sessions seem to know about each other, but somehow the message is > not going trough. > I tried this on another solaris machine running mathematica 4 and the example > works ok. The copy I am running is mathematica 3. Is this a bug on math 3? > Or maybe solaris is configured in such a way to prevent this? I don't know > how ports work, but maybe they can be disabled by some security feature. > Will a port go trough a fire-wall? > What can I do to figure out why the message is not going trough? > > Thanks, > > -Mario