Progress and cancelling sort-selects (OpenInsight 32-Bit)
At 19 MAY 2011 11:56:37AM David Salvesen wrote:
I can see that there has been some discussion about cancelling sort-select statements in the past, but have not been able to find if this is yet possible in recent versions of OI.
On a similar note, is it possible to get progress information back from a sort-select and other routines like, for instance, lh_verify? I notice that there is progress provided on the TCL, but this would appear to only be when records are being prepared for OIPI and not during the sort-select stage.
Many thanks
David
At 19 MAY 2011 06:04PM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
This is not intended to dissuade anyone else from answering but our next blog post will be on exactly this subject and is due in a few days.
World leaders in all things RevSoft
At 20 MAY 2011 09:27AM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Just to elaborate on the original post, the reason for this is that we've had to do just that very thing, and are writing some bits ourselves, which is why we have a blog post ready and under construction.
World leaders in all things RevSoft
At 20 MAY 2011 11:18AM John Bouley wrote:
Is this technique similar to the one that has been suggested in the past? By using a symbolic that has a stored procedure attached you can maintain some sort of progress information. I never liked this approach as it required adding the field onto every select you wanted to add progress.
Ultimately, the solution would be for Rlist to output some information regarding what row it is on and how may rows, possibly in a global variable. Then provide a a means to hook into the resolve process. This way a programmer could read the status information and update his own progress…
John
At 20 MAY 2011 11:25AM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Ultimately you are going to have to piggy back something onto the reduce. If you don't like adding a symbolic then you can simply shell the RLIST call. What we are using is an RLIST compatible routine (PLIST if you will though in our case GEN_SELECT_WITH_PROGRESS) which performs all of the heavy lifting meaning the programmer just calls this in place of RLIST.
World leaders in all things RevSoft
At 23 MAY 2011 08:02AM David Salvesen wrote:
Thank you Andrew. I look forward to seeing your blog.
BTW, can you confirm that this will also deal with cancelling, or is it just progress.
In a similar vein, could the progress method be cross-applied to the LH programs (fix, verify) too?
At 23 MAY 2011 09:48PM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Blog is just undergoing final proof checking. It can deal with cancelling too though given your way of working it might take a little extra work. I'd have to check the other programs - certainly works with index rebuilding!
World leaders in all things RevSoft
At 24 MAY 2011 10:13AM Don Muskopf wrote:
Would your SENL article on Test_Abort apply here? I quess it depends on whether the SELECT routines call Test_Abort.
At 24 MAY 2011 10:34AM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Yes the TEST_ABORT stuff applies to the btrees (that's a WIP in progress in house fitted in around real work) but the cancel select is down to the added in line symbolic.
World leaders in all things RevSoft
At 24 MAY 2011 12:19PM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Blog article posted.
World leaders in all things RevSoft
At 25 MAY 2011 12:51PM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
To be more accurate "unedited blog article posted". Due to a SNAFU we posted an old version of the blog article, this has now been corrected.
World leaders in all things RevSoft
At 01 JUN 2011 01:00PM David Salvesen wrote:
Thank you!
At 02 JUN 2011 06:32AM David Salvesen wrote:
Sorry, another question! The method is pretty close to what we were thinking and we do already use our own front-end to RLIST processing so that is good.
My question is: how would you do the cancel with this method? In case it is relevant, we won't be using an OI front end for the progress and cancellation, but hopefully the hook/method is still available outside of OI.
Many thanks
David
At 02 JUN 2011 09:05AM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Basically you can use any one of a number of techniques. Perhaps the easiest would be to display a dialog with a cancel button and if the cancel button is clicked then close the dialog. Within the progress logic check for the existence of the handle for the dialog and if it disappears then issue an "Abort all".
World leaders in all things RevSoft