Bug 234703 - [Scanner Discovery] "Import build log" should be more fault tolerant on file paths
Summary: [Scanner Discovery] "Import build log" should be more fault tolerant on file ...
Status: ASSIGNED
Alias: None
Product: CDT
Classification: Tools
Component: cdt-build (show other bugs)
Version: 5.0   Edit
Hardware: Sun Solaris-GTK
: P3 enhancement (vote)
Target Milestone: ---   Edit
Assignee: cdt-build-inbox@eclipse.org CLA
QA Contact: Jonah Graham CLA
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-05-29 15:16 EDT by Martin Oberhuber CLA
Modified: 2020-09-04 15:20 EDT (History)
2 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Martin Oberhuber CLA 2008-05-29 15:16:42 EDT
Consider following use case:

* An admin sets up a project on host A, builds the project, imports build log
  and sets up static analysis
* Other team members should use "best possible" static analysis settings 
  (i.e. per file discovery), but don't build the entire project themselves 
  because it is too huge

One possible solution in this case is, that the admin shares the build log on some network share, or checks it into version control along with the project. During setup, each team member imports the build log to get the good "per file" settings.

Problem with this setup today is, that when analyzing the build log, admin likely compiled the project in different file system locations than the team members have their workspaces in. Therefore, the paths from the build log cannot be matched against what team members have in their workspaces.

The discovery mechanism should be more fault tolerant here, e.g. matching file names against the workspace and checking only partial relative paths of the file from the build log. Or, allow some path substitution rules.

Another option for this setup would be that a pre-built index (PDOM) not only includes the symbols, but also the macro / include path information that was used to generate the symbols for each file. Then, if individual users have to re-build index for one particular file (because they changed it), configuration information for that file can be taken out of the PDOM.