Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
AW: AW: AW: AW: [geclipse-dev] Filesystem bug(s)

Hi All,

 

Just to clarify why Harald called it a temporary solution. The fix works for non-empty directories only. Empty directories will not be refreshed when they are opened the next time, at least if their parent was not refreshed before. So assuming a user opens a directory that is empty, closes it again without closing the parent folder, a new file appears in that directory, the user re-opens the directory he will not find that new file unless he at least collapsed the parent folder forcing a refresh of the child folders.

 

Nevertheless I think this is the best solution we currently have …

 

Thanks Harald,

 

Mathias

 

Von: geclipse-dev-bounces@xxxxxxxxxxx [mailto:geclipse-dev-bounces@xxxxxxxxxxx] Im Auftrag von Harald Gjermundrod
Gesendet: Donnerstag, 28. Februar 2008 09:16
An: Developer mailing list
Betreff: Re: AW: AW: AW: [geclipse-dev] Filesystem bug(s)

 

Hi All, 

 

After discussing with Mathias I have committed the fix below as a temporary solution. Please test out the tree in the Grid Project View and report any problems. 

 

Thanks,

Harald

 

On Feb 27, 2008, at 5:49 PM, Harald Gjermundrod wrote:



Hi Mahias

 

I can help you tomorrow. I have tried with the following code

==

      //this.treeViewer.setChildCount( container, 0 );

      //this.treeViewer.setChildCount( container, size );

      container.dispose();

      container.setDirty();

==

 

and it seems to work on Linux and Mac. I didn't want to commit it until you have taken a look at it.

 

Thanks,

Harald

 

On Feb 27, 2008, at 4:58 PM, Stuempert, Mathias IWR wrote:



Hi All,

I have committed another bunch of fixes that ...

1) resolve the problem of empty folders. You should see now a message replacing the progress bar and mentioning that the folder is empty.

2) introduces proper progress monitoring for the info system. Therefore of course the info system has to handle the progress monitoring on its side correctly. Nick, maybe you have a look at this.

The fixes are spread among various plug-ins, so to finally get them I recommend synchronizing your whole workspace. As always any feedback is welcome.

Concerning the Mac OS bug I reduced the setChildCount()-calls to lazy containers rather than to all containers. So at least it should cause no problems for "normal" folders anymore. For finding a final solution I definitely need the help of a Mac OS user. So maybe Harald or Ashish can help me with this tomorrow. Unfortunately there is no time left for me today to do some more work for this.

BTW., the reason why the old child fetching routines were replaced by the new ones was to remove the locks in the workspace. This means that you are able to browse different connections in parallel. In principle any other resource action (like new file/folder or so) not conflicting with the child fetching should also be available during child fetching. So if you are testing you should also try to open several connections in parallel. This may cause problems within the same connection (e.g. opening two folders in the same gftp connection) which then is related to the EFS implementation (e.g. gftp) and should not be reported to me :)

Cheers, Mathias

-----Ursprüngliche Nachricht-----
Von: geclipse-dev-bounces@xxxxxxxxxxx [mailto:geclipse-dev-bounces@xxxxxxxxxxx] Im Auftrag von Ashish Thandavan
Gesendet: Mittwoch, 27. Februar 2008 15:26
An: Developer mailing list
Betreff: Re: AW: AW: [geclipse-dev] Filesystem bug(s)

Oops, sorry Harald, I wrongly attributed the fix to Eclipse! I just
tried this morning's EPP build on the Mac and I have the same problem as
those many months ago. If it is any help at all, I'm given a message:
'Invalid memory access of location 0000012a eip=90820656'

Regards,
Ash

Stuempert, Mathias IWR wrote:

Hi Ashish,

 

I seem to remember discussing this with Mathias at GKS'07 in

Karlsruhe.

Since then, there has been an update to Eclipse which fixed this

problem

on Mac OS X (Tiger). Have any major changes to this code been done

recently?

 

Months ago Harald G. did a fix in the code that made it working on Mac

OS. Nevertheless I had to revoke this fix since it caused the whole

system to fetch children of virtual nodes when they are collapsed rather

than when they are expanded. Therefore it is now not working any more on

Mac OS but on the other platforms.

 

Cheers, Mathias

_______________________________________________

geclipse-dev mailing list

geclipse-dev@xxxxxxxxxxx

https://dev.eclipse.org/mailman/listinfo/geclipse-dev


--
--
Ashish
----------------------------
Ashish Thandavan
ACET Center,
School of Systems Engineering,
Philip Lyle Building,
University of Reading,
Whiteknights Campus,
Reading RG6 6BX
**********************
Tel:  +44 (0) 118 378 7645
Fax: +44 (0) 118 378 5224
---------------------------
_______________________________________________
geclipse-dev mailing list
geclipse-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/geclipse-dev
_______________________________________________
geclipse-dev mailing list
geclipse-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/geclipse-dev

 

 

 

 

 

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

 

Thanks,

Harald Gjermundrod

 

 

 

 

Back to the top