6 replies to this topic

[wicoulte]

Hey,

We got an error 010 today - 'Function excedes workspace'. The doco says this means the function is too big for the 'task buffer' and the function needs to be split into multiple functions. This particular function has been in use for years. Is this task buffer associated with the user process executing the pro exececutable? Something else? Can it be configured?

We're running pro-iv 4.0 on Tru64 unix. Can we effectively increase the user's memory using the ulimit command?

Thanks,
Bill Coulter

[sebago62]

Hey,

We got an error 010 today - 'Function excedes workspace'. The doco says this means the function is too big for the 'task buffer' and the function needs to be split into multiple functions. This particular function has been in use for years. Is this task buffer associated with the user process executing the pro exececutable? Something else? Can it be configured?

We're running pro-iv 4.0 on Tru64 unix. Can we effectively increase the user's memory using the ulimit command?

Thanks,
Bill Coulter

Hey Bill,

I've been using PRO IV since version 3. something and am now using 5.5

We have never been able to get around that message, though it's always happened to a developer for us, not a user. We've usually had to split the function up, though sometimes we've been able to re-work a large logic ID to get thru it. But one we get that error message the function is hosed till we do something about it.

Hope you have better luck.

Tom

[andykay]

Bill,

As Tom stated, there is no way to bump up your system to get rid of this error. The only workaround we've been able to do, is to either split the function (Ie. Make a pre-existing non-global LSCALL global), or use pre-existing logics multiple times throughout the function if you get the dreaded No More Logic Ids Available error message, and to remove multiple lines of the same code in multiple Logic Ids that adds to the size of the function.


Ie. - Look at your ARE logics. Usually, these will contain just a DSEL, or a DSELF command. Now, look to see if there are multiple different logic ids that can be removed, so you'd use just one logic id multiple times and thereby reduce your function size.


Good Luck,

AK

[Rory]

Hi Bill

We have sometimes had this error after adding some fields to a table and then re-genning a function that uses the table, I don't know if this is relevant to your scenario. We got around it by putting the read of the table into a global function and just passing the required fields.

Rory

[wicoulte]

Hi Bill

We have sometimes had this error after adding some fields to a table and then re-genning a function that uses the table, I don't know if this is relevant to your scenario. We got around it by putting the read of the table into a global function and just passing the required fields.

Rory

Thanks to all or the advice.

We haven't done anything and the problem has gone away. The function is running in production as normal. Go figure...

Bill

[Los Rouse]

We occasionally hit this limitation with Mainframe PRO IV.

Try checking logic zero for large scratch array(s) and, if possible, reduce their size. This has been a ‘quick win’ in the past until we have had time to break the function down.

[Chris Pepper]

Confusingly the FUNCTION EXCEEDS WORKSPACE message gets issued in two completely different circumstances.

One is during a Gen - this will happen consistently (although the actual size of the Function where this will happen will vary between different platforms and different versions of ProIV. This is where rewriting the Function may be necessary.

The second is when a Function is running (as in the original post). The Function runs out of stack space. You can demonstrate this by doing a recursive LSCALL. Things that can cause this are recursive calls (LSCALL, Global Logic, FLD statements). Whether these actually trigger the problem may depend on the data, or on how the Function is written. In the original problem something in the data caused the Function to use more stack space than expected and triggered the problem - this doesn't happen every time.

I'm sure I've posted about this on here before.