RE: [cdt-core-dev] CDT Indexing

[Chris Recoskie <recoskie@xxxxxxxxxx>]

As Doug says the index is larger than the source. If you are going to copy the entire index over to the local machine you are better off copying the entire source to the local machine and indexing it locally, because the source is smaller. I think that much Doug and I agree on.

The approach we're working on for PTP-RDT is that the index is generated on the remote machine, and there is more or less what you would call a query server that sits on top of it. On the local machine you make query requests to the remote machine and it gives you the results you are interested in. We have hooked this up to the CDT indexing framework via the appropriate extension points so that resource changes in an RSE project hosted over EFS are automatically communicated to the server so that it can index the delta and keep the index up to date. This sounds to be more or less what you are asking for.

Our approach does currently require using a CDT project, although the server code doesn't know anything about CDT projects, so you could in theory write a front end for it on the local machine that only used the RSE UI (this is what our prototype last year did). In our opinion the CDT project explorer UI gives a richer experience though.

===========================

Chris Recoskie
Team Lead, IBM CDT Team
IBM Toronto
http://www.eclipse.org/cdt

vrushali babar <vrush_com@xxxxxxxxx>

vrushali babar <vrush_com@xxxxxxxxx>
Sent by: cdt-core-dev-bounces@xxxxxxxxxxx

05/29/2008 01:34 AM

Please respond to
"CDT Core developers list." <cdt-core-dev@xxxxxxxxxxx>

To	"CDT Core developers list." <cdt-core-dev@xxxxxxxxxxx>
cc
Subject	RE: [cdt-core-dev] CDT Indexing

Hi Doug,
 
Sorry for an ambiguous reply. Let me try once again. 
Since I have extended RSE (Remote System Explorer) and integrated local environment; Local machine acts just as an interface. Every process here runs on remote machine.
I don’t maintain local copy of any resource so those resources can not be indexed by CDT so the approach I thought of –
 
- Indexing all the files on server and copying the generated files by indexer to local machine.
 
Server maintains the whole copy of the project and developer checks out few files as per the requirement and go for incremental build after the changes.
Again whatever files he checks out gets displayed in RSE’s System View which is not linked with CDT so those files can not be indexed.
 
So do you see any way through which I can achieve indexing of files, displayed in RSE’s System View? 
Does CDT provides any extension point through which I can make use of CDT indexing for only those files in which is he making changes?
 
Thanks,
Vrushali. 

"Schaefer, Doug" <Doug.Schaefer@xxxxxxxxxxxxx> wrote: 
Are you using the CDT to make those changes? The CDT reparses files as they change and adds the changes to the index. Remote update should be triggering resource change events for the files that change or lots of things don't work.
 
Also, I'm not sure what you mean by user does not maintain the whole copy of the project. What files show up in the Project Explorer? Everything that shows up there is indexed.
 
Doug.

---

From: cdt-core-dev-bounces@xxxxxxxxxxx [mailto:cdt-core-dev-bounces@xxxxxxxxxxx] On Behalf Of vrushali babar
Sent: Wednesday, May 28, 2008 10:43 AM
To: CDT Core developers list.
Subject: RE: [cdt-core-dev] CDT Indexing

Hi Doug,
 
Thanks for your reply.
Since indexed file is larger than the original file, i am fine with transfering it to local only once and referring it but what about the changes that i am doing in the remote file. How will I reflect those changes in the index file so that I shall get better search results even by considering just made changes.
Also, user does not maintain the whole copy of project. He just checks out the required files so how can we handle indexing only for those files?
Could you please suggest anything on these points?
 
Thanks,
Vrushali.

"Schaefer, Doug" <Doug.Schaefer@xxxxxxxxxxxxx> wrote:

This has been discussed often in the last year or so. My conjecture is that the size of the index is actually larger than the size of the source. And since the fast indexer only parses each file once, it's actually faster to keep the indexer local and access the files remotely.

But others have a different opinion.

Doug.



---

From: cdt-core-dev-bounces@xxxxxxxxxxx [mailto:cdt-core-dev-bounces@xxxxxxxxxxx] On Behalf Of vrushali babar
Sent: Wednesday, May 28, 2008 8:46 AM
To: cdt-core-dev@xxxxxxxxxxx
Subject: [cdt-core-dev] CDT Indexing

Hi,

CDT provides efficient indexing for local files.is there any way to achieve that indexing for remote files?
if CDT stores those indexed files somewhere and refers it to them later then how about a following approach - - having eclipse on remote server
- indexing all the files on remote server with Eclipse CDT
- copy those generated indexed files from remote server and store them locally
- refer these files in case of any search demand.

Please correct if i am going wrong some where or if there exists any better option to which i can shift to.

Thanks,
vrushali.
_______________________________________________
cdt-core-dev mailing list
cdt-core-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cdt-core-dev

_______________________________________________
cdt-core-dev mailing list
cdt-core-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cdt-core-dev

_______________________________________________
cdt-core-dev mailing list
cdt-core-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/cdt-core-dev