PDSOE 11.4 'hangs' for several minutes after compile

Posted by Rom Elwell on 07-Apr-2015 08:25

Setup:  PDSOE 11.4 running on a Windows Ultimate VM

Issue: A coworker is observing that when compiling/building source files, the IDE becomes unresponsive for up to three minutes immediately after the compiler reports an error in a source file. We have tested several use cases for this anomaly; removing a period (.) from the end of a statement in a procedure (*.p), a class (*.cls), and an include (*.i).  He will save the file in question and initiate a compile (of the file) or build (of the target project). His IDE responds identically in every use-case.  The compiler reports the error and the IDE 'hangs', becoming unresponsive for some excessive amount of time afterward.  The IDE 'reports' that it is still building the offending file during this unresponsive time.

I am unable to reproduce this issue in my development environment, so, we are targeting the root cause as a possible issue with his development environment (possible configuration issue?)

Question: Is there documentation that will help me learn the workflow PDSOE uses when the user initiates a 'compile' action?  I'm eager to learn what actions occur after the compiler is called.

Thanks!

All Replies

Posted by Swathi Yellavaram on 08-Apr-2015 09:27

Hi,

There is no specific document available showing what pdsoe compile operation does. If we just compile one file - it does compile complete file and show all the errors in file. It updates errors information in problems view and show error markers in file.

Can you please let us know if the issue of hang is seen for any file in the system or its specific to a file?

Thanks,

Swathi.

Posted by sudhan on 09-Apr-2015 01:42

Hi Rom

In case it is specific to a single file, would it be possible to share that with us? And any other details you may have about the number of open projects, number of files in each? The behavior you observed could be a result of an issue reported in 11.4 where any change to a file triggered a lot of I/O operations by the framework. This has been optimized in 11.5.1. To check if this is due to the same bug, can you please try the same with just a single project containing just a few files (the bug manifests itself when there are large amounts of files). Based on this and if you find the issue interfering with your usage of PDSOE, we can then issue a hotfix once the request is raised for 11.4.

Regards

Sudhan

Posted by Rom Elwell on 09-Apr-2015 06:46

Thank you for the generous feedback. The issue manifests irregardless of the file type that contains the error.  In the simplest terms, any error in a *.i, *.p, or *.cls file that is caught by the compiler causes the subsequent 2-3 minute unresponsiveness within PDSOE.  We've ensured that 'automatic publishing' is disabled for his Mobile projects (thinking that this might have been the root cause for the unresponsiveness...it was not).

Is there any way to gain visibility into the operations of the associated prowin32.exe or javaw.exe during this period of unresponsiveness?

Thanks!

Posted by Matt Baker on 09-Apr-2015 07:58

 
Is the compiler actually still compiling the file?  Can you check cpu usage of prowin executable and eclipse.exe?  There have been oddball cases in the past with very deeply nested structures making both OEM and the database take an absurdly long time to process the file.
 
 
 
[collapse]
From: sudhan [mailto:bounce-sudhan@community.progress.com]
Sent: Thursday, April 09, 2015 2:43 AM
To: TU.OE.Development@community.progress.com
Subject: RE: [Technical Users - OE Development] PDSOE 11.4 'hangs' for several minutes after compiler reports an error in source file
 
Reply by sudhan

Hi Rom

In case it is specific to a single file, would it be possible to share that with us? And any other details you may have about the number of open projects, number of files in each? The behavior you observed could be a result of an issue reported in 11.4 where any change to a file triggered a lot of I/O operations by the framework. This has been optimized in 11.5.1. To check if this is due to the same bug, can you please try the same with just a single project containing just a few files (the bug manifests itself when there are large amounts of files). Based on this and if you find the issue interfering with your usage of PDSOE, we can then issue a hotfix once the request is raised for 11.4.

Regards

Sudhan

Stop receiving emails on this subject.

Flag this post as spam/abuse.

[/collapse]

Posted by jquerijero on 04-Nov-2015 14:22

In my case, compile finishes comparable to desktop RoundTable when there is no error. It hangs for several minutes if there is an error. I'm not sure if it is compiling still or it's just doing some other things. I ended up doing my initial compile in desktop RoundTable first which is faster in this case.

Posted by Blake Stanford on 04-Nov-2015 15:08

I have found that some code will not compile with both xref and listing turned on.  It manifests as the compile just hanging.  If you have your Roundtable TSMS Builder Options - listing on you might try turning it off.  I think there is a KB entry out there referencing the listing option.

Posted by Jeff Ledbetter on 05-Nov-2015 01:11

Hi.
 
Is this a single Object compile or many Objects? Objects are compiled in batches. After each batch is compiled, it is then xref’d (on the server). However, it should not take long enough to appear to be hanging though.
 
Feel free to contact us at support@roundtable-software.com for assistance.
 
Jeff Ledbetter
skype: jeff.ledbetter
 

Posted by Jeff Ledbetter on 05-Nov-2015 01:13

Hi.
 
I’ve not seen that occur before.  Server or Client compile?  Any application error in the appserver log?
 
The end result is just a COMPILE statement being issued so I can’t imagine why it would hang unless something else is going wrong.
 
Jeff Ledbetter
skype: jeff.ledbetter
 

Posted by Blake Stanford on 05-Nov-2015 08:25

I worked with Randall from  RTB support some time back.  He had me turn on some logging for RTB and

Eclipse that helped with figuring out the compile problem I had.  He had some very good insight.

Joe do you have your workspace and projects on network drives?  I moved my workspace to my  local drive but left my project files on the network drive.  That helped with performance, as Eclipse continually performs workspace file updates.    

Posted by Frank Meulblok on 05-Nov-2015 11:02

Another "gotcha" to keep in mind:

Make sure you have more than one CPU core available.

Eclipse in general really doesn't run well in single-core environments - it runs a lot of background activity on seperate threads to keep things moving along, but if all those threads still have to queue on the same core that just can't work well.

This is more of an issue if you're running in Virtual Machines, where you have to provision a number of cores for the VM.

(When it comes to physical hardware, it takes real effort to find a single-core machine that's less than a decade old. )

This thread is closed