On Dec 7, 2010, at 4:46 AM, Brian McIlwrath wrote:
> Hi David,
>
> I am no longer sure that I need to look that hard for HDS problems!
>
It seems like all the problems were actually higher up in the code. That's a bit embarrassing although it seems what has really happened is that we have relearnt that HDS temp files keep hold of their content. I think we were all expecting that annulling a temp locator would erase all the structures attached to that locator but now it seems that temp locators are just a way of storing temporary stuff until the process shuts down.
> The temporary HDS file (at least for COLLAPSER) is remaining pretty
> static in size now! I will see what Malcolm can break!
>
I think you made some fixes to temp files along the way. Can we have those? Thanks for looking into things for us.
> The problem I can see at times in HDS is the temp file growing
> uncontrollably after a certain point (as you saw yesterday). Rodney's
> comment would indeed be rather hard to debug!
We are seeing a very strange mapping problem with HDS that has turned up sometime since September which we are trying to debug (but it can't be an HDS problem as such because it worked in August).
--
Tim Jenness
Joint Astronomy Centre
|