I have managed to get a C Autotools hello world project working remotely
using changes I made to the rdt branch and a hacked up CDT. To create
the C project, I simply created a C Autotools Hello World project and
specified an RSE location in the wizard. I selected a directory on my
remote machine. I had to manually add in the project name to the chosen
location (i.e. the project directory didn't exist already so I chose a
workspace I had lying around). I was careful not to screw up the
connection name which was added to the URI as a query.
After that, the project was created on the remote machine. Note that I
have a CDT branch with some hacks in it to support remote projects (I
gave the CDT its own version of the remote proxy manager and its own
RDTProxyManager and then I fixed up some key locations in the CDT code
as I found them).
With the changes, an Autotools project configures, runs make remotely,
and builds a binary which shows up in the Binaries folder. I am able to
run the binary as a Local C/C++ executable (it doesn't know it is
remote) and the results are shown in the Console. There are some extra
lines added in the console output regarding shell completion codes, but
they are only a minor nuisance at the moment and may be because I added
Dstore tracing on the server. The indexer works, but there is a bug in
the version of the indexer/codan I have in my CDT snapshot whereby it
claims not to know about the value of EXIT_SUCCESS. The same thing
happens for a local Managed Build C project so it isn't a remote issue.
If I click on the EXIT_SUCCESS macro in the hello world program and
hit F3, it finds the stdlib header file declaration. I did some work to
make sure it was getting the right TranslationUnit, but I have not fixed
everything so header paths, etc.. may still be local. Just more fixes
to be done in the CDT.
Anyway, I feel it definitely proves the concept. I have checked in my
latest changes into the rdt branch for the linux tool pieces. I still
need to make the hacked CDT branch available today.
I had a lot of trouble getting the rseserver to work. It doesn't
support kerberos which we use here and the ssl stuff failed on me so I
had to hack the perl scripts a bit. I am currently investigating with
the help of some RSE folks regarding some of my issues. I hope to have
some documentation to add on how to set it up properly. The rseserver
docs aren't very detailed. SSH tunnelling is something else I want to
look at. It doesn't work because you must specify the host as LOCALHOST
and the connection is being ignored at this point so the project gets
created locally. I couldn't get the Dstore connection to work if I
specified the port with the host name (e.g. LOCALHOST:21769).
-- Jeff J.
_______________________________________________
linuxtools-dev mailing list
[email protected]
https://dev.eclipse.org/mailman/listinfo/linuxtools-dev
