[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cdt-dev] First gotcha with add/exclude children of FFS

Title: Re: [cdt-dev] First gotcha with add/exclude children of FFS
Doug,

Can we get together at EclipseCon to discuss this issue specifically? Do you know the right platform people to invite to the meeting? We really need to piece together a plan.

I’m sure you have enough to do so if you can tell me who to recruit I can help organize the meeting.

Thanks - Ken


From: "ext Schaefer, Doug" <Doug.Schaefer@xxxxxxxxxxxxx>
Reply-To: "CDT General developers list." <cdt-dev@xxxxxxxxxxx>
Date: Wed, 20 Feb 2008 18:44:57 -0800
To: "CDT General developers list." <cdt-dev@xxxxxxxxxxx>
Subject: RE: [cdt-dev] First gotcha with add/exclude children of FFS

So, you know. The more I think about what you guys are saying, I'm realizing that the EFS solution probably is the right one. The objective should be to turn the IResource tree into a logical project view and to remove all notions that it represents physical file layout. That unfortunately starts with the .project and .cproject files, but I think there are tricks we can do there. The .settings may be harder but let me sleep on that.

At any rate, this has piqued my interest again and I'll work on reviving it and see where it goes. I'll try to get a prototype working by EclipseCon so we can talk about it more concretely.

Cheers,
Doug


From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Brunauer, Walter
Sent: Monday, February 18, 2008 8:53 AM
To: CDT General developers list.
Subject: RE: [cdt-dev] First gotcha with add/exclude children of FFS

Hi Warren,

well, the confusion my origin from the different meanings of what project setup is: for me, project setup is not equal to build setup. I.e., on our projects, the build setup is an independent step from the project setup. We intentionally separated this to overcome all kind of issues you obviously experienced as well. And this is how we see it:

1. Create a project at the desired location (everything beneath this root location is part of the project, but it can be an empty project just as well with linked resources added to it later). By default, the build setup is identical to the project content (there is one build target, linking/archiving everything together).

2. If (a) specific build setup(s) are needed, it is possible to specify as many build targets with arbitrary contents as desired. This approach separates the physical file system layout from logical build layouts, and it even works beyond project boundaries. IOW, no matter from where source files are pulled in (the same projects, nested projects, outer projects, sibling projects), one is able to specify any build setup exactly are needed, as long as all source files are known to Eclipse (as resources).

HTH,

Walter


 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of  Warren.Paul@xxxxxxxxx
Sent: Freitag, 15. Februar 2008  14:51
To: cdt-dev@xxxxxxxxxxx
Subject: RE: [cdt-dev] First  gotcha with add/exclude children of FFS

 
 
Hi Walter,

 
 
I forgot all about the absolute paths issue with linked  resources.  I'll update the wiki.

 
 
I'm a bit confused about your comment about this not  being a project setup issue.  We have our own builder as well, and it  will happily build whatever the build description says, whether those files  are under the project root or not.  We even have our own project explorer  view which shows a logical representation of the project rather than the  physical file system layout.  But we still run into a lot of issues when  files are not under the project root - that is, when you can't get an IFile  for them.

 
 
We have a wide range of user types from small application  developers to large system developers.  In many cases, a users code base  consists of hundreds of directories with thousands of source files.  In  such a source base, there are many hundreds of build artifacts and almost as  many "logical projects".  It is a huge problem for these users to be able  to create projects currently.  They will typically have a few projects  going at a time, but many times the natural project root for all of them will  be the same.  We've found ways to work around some of the other issues,  but not this one.  It sounds like perhaps you guys have found a  way.  Could you elaborate on how you setup your  projects?

 
 
Thanks,
Warren

 
 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of ext Brunauer,  Walter
Sent: Friday, February 15, 2008 1:56 AM
To: CDT  General developers list.
Subject: RE: [cdt-dev] First gotcha with  add/exclude children of FFS

 
 
Hi Warren,

 
 
FWIW, you did not mention anything about linked  resources and absolute paths these persist in the .project file by default.  Again a big issue around linked resources in combination with sharing project  within a team (even without team support), and one more reason why they appear  to be so cumbersome to handle. To me it seems many times one has  to unsell linked resources to users: Whereas linked resources  are (kind of!) nice for evaluation purposes (because, yes, in this case  you might not want to pollute your sources), as soon as you start serious  development, you run into all kind of troubles. The hurdle to get everything  right from the beginning is overwhelming for novices (e.g. its not possible to  change a linked resource to use a variable later). Sorry, I don't know how to  add this to the Wiki page...

 
 
Having said that, the scenario you describe is  really about having the flexibility around build and indexer setup, not around  project setup, IMO.

 
 
It's rather classic: users have common  code they want to reuse in multiple applications - so they create one or  a set of libraries out of it, within one or a set of projects. Of course,  indexing should be able to handle only code going into these libraries, and  optionally ignore the rest. Then, they create their application projects, which  use the binary artifacts of the library project(s). Now it would be great if  they would have automatic support for application linkage specification, i.e.  some nice wizard or UI allowing to select the library binaries of other  projects to be linked in, without the need to specify it manually in the  linker options. And probably also desired: during application code  development, the public API's of all used library projects should be the only  thing they see WRT code completion, etc. I guess, some UI would be needed for  this as well.

 
 
And now think of all developers in the world.  Wouldn't it be great to give as many of them the freedom to choose how to  achieve this? Either everything in one project, or one project per build  artifact, or one project per module/application/product, or with nested  projects... its possible. Our commercial IDE based on CDT supports all this,  and we did not have to provide some EFS or work with linked resources.  Well, we had to override the build system, and this is IMO the place to solve  this in CDT as well.

 
 
 
Again, I don't see anything specific to project  setup. The issue around having the source tree polluted with project files - I  don't think this is the big thing. I would not leave the Eclipse path in this  area at all and allow to separate the project file from the project location.  Its a very general paradigm of Eclipse, and I am pretty sure doing everything  differently will generate lots and lots of problems in all kind of areas  (probably much more than you already identified), unless you make it a new  Eclipse way (add/change this in the platform, not in CDT, that  is).

 
 
Just my 2 or 3 cents again,

 
 
Walter

 

 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of  Warren.Paul@xxxxxxxxx
Sent: Donnerstag, 14. Februar 2008  23:35
To: cdt-dev@xxxxxxxxxxx
Subject: RE: [cdt-dev]  First gotcha with add/exclude children of FFS

 
 
I've updated the Wiki page http://wiki.eclipse.org/CDT:Flexible_Project_Structure with  some more thoughts on the issue.  It would be great to get feedback  from other CDT users - both those shipping C/C++ IDE's and end users.   You'll see that I'm not convinced that the linked resources route is a  viable option.  Maybe we can get the platform team involved in the  discussion to help find the best route forward.

 
 
Thanks,
Warren

 

 
 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of ext Schaefer,  Doug
Sent: Friday, January 25, 2008 11:15 AM
To: CDT  General developers list.
Subject: RE: [cdt-dev] First gotcha with  add/exclude children of FFS

 
 
I guess what my investigation has shown me that the EFS  solution and linked resources are pretty much identical. I really  noticed this when trying to figure out how to persist the adds and found  myself wishing I could add that to the .project file just like linked  resource are. And they are....

 
 
I think all the issues that we have with linked resources would be equally as bad with the EFS solution,  possibly worse because the EFS adds are hidden. The CVS one is a great  example. I really doubt CVS would work with the EFS solution either. And I  don't want us to think EFS would be better since it's not in the platform  where we'll have a battle getting changes. Any platform changes required to  make linked resource work correctly would also need to be done for  EFS.

 
 
So my hope is to save the effort at implementing  the add/remove functionality since I believe that's already there with  linked/hidden resources. We can then focus on making linked resources  work where we need them and improving the workflows. But this  really needs to start now.

 
 
So, Warren, you've somewhat started a list of  workflows that we'd like to support with this solution. This is a great  place to start. I've created a Wiki page to start capturing these. Please  feel free to add more information, especially to the workflow section. When  we have that we may get a better idea of which of the two solutions will  work best.

 
 
   http://wiki.eclipse.org/CDT:Flexible_Project_Structure

 
 
Thanks,

Doug


 
 
 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of  Warren.Paul@xxxxxxxxx
Sent: Friday, January 25, 2008 11:22  AM
To: cdt-dev@xxxxxxxxxxx
Subject: RE: [cdt-dev] First  gotcha with add/exclude children of FFS

 
 
We've been working on Eclipse/CDT based products for  about three years now.  I'm sorry to say that the project model is  still not satisfactory for our purposes.  We've tried many angles, but  are still stuck with some pretty serious limitations.  I've volunteered  to investigate the EFS route to see if it will help at all.  Based on  this thread I'm assuming it won't.

 
 
Let me give you a brief overview of how our users work,  and then discuss the problem we've run into.  I don't think any of this  is specific to our users BTW.

 
 
Most of our users have existing code bases.  They  simply want to "import" it into the IDE.  Others will create new  projects from our templates.  The new projects are created in the  workspace.  Imported projects could be anywhere in the file  system.  Often times they will import several projects from the same  source tree.  This is where our biggest problem is.  Let's say the  source base looks like this:

 
 
C:\MyProjects\Project1\...


C:\MyProjects\Project2\...
 
 
C:\MyProjects\Common\...
 
 
 
Because both projects  share code in the Common directory, the logical root project directory for  both Project1 and Project 2 is C:\MyProjects\.  But in Eclipse you  can't have two projects with the same root.  This is where the .project  and .cproject files are created.  So currently our users would import  Project1 with the natural root (C:\MyProjects\), but Project2 has to be  rooted at C:\MyProjects\Project2\.  This means that any source/headers  from the common directory are not under Project2.  This means those  files are not in the project explorer for that project, are not indexed,  etc..  We logged this against the platform - https://bugs.eclipse.org/bugs/show_bug.cgi?id=78438.   Basically if you put the .project file anywhere, but have a project root  attribute, this would cease to be a  problem.
 
 
 
Our first product actually always created the  .project in the workspace, and for imported projects, would create links to  files and folders.  We ran into so many issues with this that we had to  change the model.  I don't recall all of the issues, but here's a list  of some:
 
 
 
- Version control simply didn't work at  all
 
 
 
- You can't make file system changes with  links.  For example, if you want to rename a file or folder, or move a  file around, you can't do this with linked resources.  It only changes  the link itself, not the underlying  resource.
 
 
 
- Creating new resources in a project with links is  confusing at best.  Let's say you have a project with a linked folder  and file at the root.  If you create a new file or folder at the root,  it is created in the workspace, not where the other folder/file are in the  file system.  But if you create a new file under the linked  folder, it gets created where you'd  expect.
 
 
 
- The location of the .project/.cproject files are  problematic.  Some users will want to keep these in version control,  while others won't.  Those that do want them created in the source  tree, but those that don't want them elsewhere, like the workspace.  I  forget now why this was a problem with linked resources, but there was  something weird going on  there.
 
 
 
I suppose it's worth noting that the last time we  really looked at this was in Eclipse 3.2, so some of this may have been  fixed by now.  But I doubt it.  In general linked resources are  second class citizens.  Some IResource API's just don't work for linked  resources.  Just search for references to IResource#isLinked for  "special handling".  I suspect that we'll run into similar issues with  EFS though - see getLocation vs  getLocationURI.
 
 
 
We also have the same issue that Doug is trying to  address (hiding some branches in a source tree).  This is much less of  an issue for us though.  You can already reduce the scope of the  indexer and the build.  The only real issue for us is for a very large  source tree, you're going to get IResource's for everything, which slows  things down quite a bit.  There is actually somewhat of a problem in  reducing the indexer scope - see https://bugs.eclipse.org/bugs/show_bug.cgi?id=178159.
 
 
 
The hidden attribute addition sounds promising for  hiding resources under the project root, but doesn't really do anything to  add flexibility to the contents of a project.  EFS sounds like it would  though.  What I mean by that is, having resources under a project that  are real resources, not linked, but that don't live under the project root  in the file system.  I've just started looking into EFS, so maybe it's  a bit of wishful thinking at this point, but I'm hoping we could create a  project anywhere, and when we create it we pass the URI location from our  own EFS.  Then when asked for the children, we could return URI's for  files from anywhere in the file system, or on other machines even.   This would seem to hold the potential for working around the issues listed  above.  We'd basically have an EFS map from what we want under a  project to the actual file  system.
 
 
 
So hopefully some of the experts can chime in  here.  Is my hope for EFS unrealistic?  Is there a different  approach we should look at?
 
 
 
Thanks,
Warren
 
 

 
 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of ext Brunauer,  Walter
Sent: Friday, January 25, 2008 1:47 AM
To: CDT  General developers list.
Subject: RE: [cdt-dev] First gotcha with  add/exclude children of FFS

 
 
Hi,

 
 
after reading this rather long thread, I'll  decided to throw in my personal opinion.

 
 
I consider this approach to work against one of  the most general Eclipse platform paradigms, where a project is defined  to be a root directory and everything in it. IMO, the more workarounds are  introduced against this paradigm, the more problems will be faced, and the  more incompatibilities (or at least unawarenesses)  created.

 
 
Isn't the whole problem you try to solve here  rather about what files should go into the build (and probably into the  indexer) than what files are part of a project? I understand that CDT has no  separation of what a project and what the build input is (well, IIRC one can  exclude specific files from the build, but in general, the project content  defines the build input, right?).

 
 
In our commercial IDE, we separated this. This not  only introduced much more powerful build setup capabilities in general, but  especially enabled users to setup build artifacts with arbitrary  contents (think of sources being compilable with different compiler flags  for different build artifacts, build input exclusion patterns, build input  from all over the workspace, multiple build artifacts within the same  project, reusable build artifacts accross project boundaries, etc., etc.,  etc.). BTW, we call this build system flexible managed build - because  that's what it is:-)

 
 
Of course, one can setup CDT projects as of today  to exactly contain what is desired (with the help of linked resources).  However, I find linked resources to be cumbersome and error prone, though  many of our customers start out with them during evaluation as well, mostly  because they are looking for a way to achieve what they did in the past with  other non-Eclipse based IDEs, but sooner or later I know of lots of  them realizing its much easier to use the features of our flexible build  system instead, especially if projects need to be shared in a team.  And now think of virtual file systems, the potential complexity of  these, hidden assumptions, restrictions, etc. Sounds worse than linked  resources to me.

 
 
I guess, the point I am trying to make is:  whatever you decide to do, make it understandable  and transparent (and of course as simple as possible to use) for  the user.

 
 
As said, just my 2 cents,

 
 
Walter

 

 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Schaefer,  Doug
Sent: Donnerstag, 24. Jänner 2008 23:17
To: CDT  General developers list.
Subject: RE: [cdt-dev] First gotcha  with add/exclude children of FFS

 
 
Jogging through the code, it really looks like the  HIDDEN feature is what I was looking for. What I haven't found yet is UI  to make a resource hidden or a navigator filter to show hidden resources  (in case you want to bring them back). Is this planned?

 
 
Assuming we have the core features available to link  in and hide resource, I think we still have workflow issues that need to  be addressed. I like Ken's idea of a file that controls the  linking/hiding. We could have an import/export mechanism for setting up  projects based on this file. A nice wizard for creating the file would  also be good, similar to the way the way the export file system wizard  works.

 
 
Given this, we may be further along than we thought  (BTW, the hidden stuff seems to have been added in 3.4  M4).

 
 
Cheers,

Doug

 
 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Schaefer,  Doug
Sent: Thursday, January 24, 2008 2:51 PM
To: CDT  General developers list.
Subject: RE: [cdt-dev] First gotcha  with add/exclude children of FFS

 
 
Thanks Michael/Szymon,

 
 
Is there a bug describing the isHidden  feature?


Doug

 
 

From: cdt-dev-bounces@xxxxxxxxxxx  [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Michael  Valenta
Sent: Thursday, January 24, 2008 11:37 AM
To:  cdt-dev@xxxxxxxxxxx
Subject: RE: [cdt-dev] First gotcha with  add/exclude children of FFS

 
Doug et al,

Szymon is really the person you want to bug  on this but I'll throw in my 2 cents ;-)  First, I have to say that a  solution at the IResource level (e.g. using linked resources and the new  hidden folder support) is infinitely better from a repository provider  perspective than an EFS based solution.  You may not get all the Team  support you want at the IResource level but a solution at the EFS level  would certainly break the existing CVS client since the CVS client isn't  EFS aware to any great extent. For instance, if you tried to hide a folder  using EFS, the CVS client would probably try and recreate it the next time  you performed a Team>Update. It is also important to note that the  Platform does not provide all the hooks required by repository providers  and I know of at least one provider that has resorted to using it's own  EFS implementation under projects that are mapped to that provider to get  the capabilities it requires. I think it is important that tooling in  Eclipse stick to using the IResource layer as the layer they operate on  and let the repository provider (or any other tooling whose responsibility  it is to manage the available files) control the underlying file system.  If there are shortcomings or enhancements required then you should push to  get them in at the IResource level.

As for the current state of  Team support for linked resources, I think the best approach is to  enumerate some specific scenarios of how you see linked resources and  exclusions working with descriptions of what you need to do today to get  Team support and what you would like to see. It is also important to know  if you expect all the links to come from the same repository (or at least  repository type) or whether a project could contain content from different  repository types (obviously the later would be more difficult to  accommodate than the former).

Hope this  helps,
Michael


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