Skip to main content

Notice: this Wiki will be going read only early in 2024 and edits will no longer be possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.

Jump to: navigation, search

Difference between revisions of "Build Enhancements"

(New page: 1) No regeneration of build.xml after changing a project's dependencies. Each project's build should ideally do whatever it takes to work before and after changes to that project's depende...)
 
Line 1: Line 1:
1) No regeneration of build.xml after changing a project's dependencies. Each project's build should ideally do whatever it takes to work before and after changes to that project's dependency list.
+
== List of current enhancement candidates ==
  
2) Dependencies are versioned. We should be able at worst to cause a build to use a specified CVS tag for all dependencies, and ideally, we should be able to specify a cvs tag for each dependency.
+
* 1) No regeneration of build.xml after changing a project's dependencies. Each project's build should ideally do whatever it takes to work before and after changes to that project's dependency list.
  
3) Single place for dependenciesRight now, we've decided to list them in manifest.mf.  Because this file doesn't allow us to control its contents, we're restricted from doing things like #2 above.  Moreover, it sounds like build.properties must also in some cases be updated with dependencies when there's a need to differentiate between build libs and runtime libs.  I think we should be updating our own dependencies.xml file, and that should be used to auto-gen, or update the manifest.mf as appropriate.
+
* 2)Dependencies are versionedWe should be able at worst to cause a build to use a specified CVS tag for all dependencies, and ideally, we should be able to specify a cvs tag for each dependency.
  
4) Any automated process should mirror manual processes.  If I pull down a project into a fresh work area and run it's build (which  automates the process of gathering dependencies), the result should look the same as if I had manually gathered all those dependencies.  Also, the fetched dependencies should remain in place and be usable at a later time.
+
* 3) Single place for dependencies.  Right now, we've decided to list them in manifest.mf.  Because this file doesn't allow us to control its contents, we're restricted from doing things like #2 above.  Moreover, it sounds like build.properties must also in some cases be updated with dependencies when there's a need to differentiate between build libs and runtime libs.  I think we should be updating our own dependencies.xml file, and that should be used to auto-gen, or update the manifest.mf as appropriate.
 +
 
 +
* 4) Any automated process should mirror manual processes.  If I pull down a project into a fresh work area and run it's build (which  automates the process of gathering dependencies), the result should look the same as if I had manually gathered all those dependencies.  Also, the fetched dependencies should remain in place and be usable at a later time.

Revision as of 17:22, 22 October 2007

List of current enhancement candidates

  • 1) No regeneration of build.xml after changing a project's dependencies. Each project's build should ideally do whatever it takes to work before and after changes to that project's dependency list.
  • 2)Dependencies are versioned. We should be able at worst to cause a build to use a specified CVS tag for all dependencies, and ideally, we should be able to specify a cvs tag for each dependency.
  • 3) Single place for dependencies. Right now, we've decided to list them in manifest.mf. Because this file doesn't allow us to control its contents, we're restricted from doing things like #2 above. Moreover, it sounds like build.properties must also in some cases be updated with dependencies when there's a need to differentiate between build libs and runtime libs. I think we should be updating our own dependencies.xml file, and that should be used to auto-gen, or update the manifest.mf as appropriate.
  • 4) Any automated process should mirror manual processes. If I pull down a project into a fresh work area and run it's build (which automates the process of gathering dependencies), the result should look the same as if I had manually gathered all those dependencies. Also, the fetched dependencies should remain in place and be usable at a later time.

Back to the top