| Summary: | [efs] cannot close project on a remote EFS host if host becomes unavailable | ||
|---|---|---|---|
| Product: | [Tools] Target Management | Reporter: | Rafael Chaves <eclipse> |
| Component: | RSE | Assignee: | dsdp.tm.rse-inbox <tm.rse-inbox> |
| Status: | NEW --- | QA Contact: | Martin Oberhuber <mober.at+eclipse> |
| Severity: | normal | ||
| Priority: | P3 | CC: | mober.at+eclipse, sptaszkiewicz |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Windows 7 | ||
| Whiteboard: | |||
|
Description
Rafael Chaves
Remote System Explorer End-User Runtime 3.1.2.R31x_v200911051655 RSE Core 3.1.1.v200907301400 RSE SSH Services 3.0.1.v200909160005 Build id: I20090916-0905 I think this belongs to Platform Resources. Closing a project should be possible even if there are errors accessing files on disk ... user intended to get rid of the project, so this should succeed or the user is locked-up. We shouldn't have any cached in-memory data at this point that can't be restored from disk at a later point. Rafael, as a workaround you should create a LOCAL project and then create a linked resource pointing to your remote files, that's the recommended workflow due to the known limitations with remote projects (especially bug 181460). That being said, I see people use remote projects again and again. It's briefly mentioned in the TM and RSE FAQ (point 5 here: http://wiki.eclipse.org/TM_and_RSE_FAQ#How_can_I_use_a_remote_workspace_over_SSH.3F ) but apparently that's not visible enough. I'd love to hear ideas how to make this more visible. Rafael, I cannot parse this : eclipse.buildId=2.3.2.201003230009-RELEASE Does this relate to Galileo SR2 (Eclipse 3.5.2) ? The problem is Project#saveMetaInfo which is a "sanity check to make sure there is something on disk before we shutdown". I think that if the file system goes offline, we could close the project anyway and just log a warning. +1 on the approach. We should make sure this isn't forgotten in 3.7 - it's a nasty usability problem and looks like there's a simple fix. Szymon P., please take a look at it. Moving to 3.8. I reviewed the code and it seems that RSE has wrong implementation of IFileStore#fetchInfo(). See javadoc for this method and compare to the changes made in bug 182006. I think that RSE should do the same thing in their code. Moving to RSE. |