Hi,
Ever had this error “The exception unknown software exception (0xc0000417) occurred in the application at 0x00a13c18”.
Solving it is easy, just erase the .AUC files in the folder:
C:\Documents and Settings\Current user\Local Settings\Application Data
Or for Windows 7 / Windows server 2008 users:
C:\Users\Current user\AppData\Local
Hi there,
here is a simple job to enable SQL tracing for all your users, this quite handy for optimizing queries. (The macro’s for modifying other fields on the UserInfo table can be found on the ClassDeclaration of the SysUserSetup form.)
boolean enable = true; UserInfo userInfo; #LOCALMACRO.FLAG_SQLTrace ( 1 << 8 ) #ENDMACRO #LOCALMACRO.FLAG_TraceInfoQueryTable ( 1 << 11 ) #ENDMACRO ; ttsbegin; while select forupdate userInfo { if(enable) { userInfo.DebugInfo = userInfo.DebugInfo | #FLAG_SQLTrace; userInfo.TraceInfo = userInfo.TraceInfo | #FLAG_TraceInfoQueryTable; userInfo.querytimeLimit = 100; } else { userInfo.DebugInfo = userInfo.DebugInfo ^ #FLAG_SQLTrace; userInfo.TraceInfo = userInfo.TraceInfo ^ #FLAG_TraceInfoQueryTable; } userInfo.update(); } ttscommit;
Make sure that client tracing is enabled in the server configuration. (Only use this in development and testing environments, tracing may affect the AOS performance.)
The results can be found in the Administration module.
Hi,
Next post will be a little tutorial on how the RunBaseBatch framework can work multithreaded. For example in the SalesFormLetter class on the method run, the following code will be found before the query iteration:
if (this.canMultiThread()) { batchHeader = BatchHeader::construct(this.parmCurrentBatch().BatchJobId); salesFormLetterEndMultiThread = SalesFormLetterEndMultiThread::newFormLetter(this, salesParmUpdate.ParmId, salesParmUpdate.Proforma); batchHeader.addRuntimeTask(salesFormLetterEndMultiThread,this.parmCurrentBatch().RecId); }
The SalesFormLetterEndMultiThread that is being created will be called when all threads connected to that bacth are processed, this will call methods like printJournal and endUpdate. Notice that all the variables that are passed in the construct method are also defined in the CurrentList macro for packing and unpacking, this is important to keep in mind when writing custom code.
In the iteration itself, another multithread batch task is created for each line.
if (batchHeader) { formLetterMultiThread = FormLetterMultiThread::newFormLetter(this); batchHeader.addRuntimeTask(formLetterMultiThread,this.parmCurrentBatch().RecId); batchHeader.addDependency(salesFormLetterEndMultiThread,formLetterMultiThread,BatchDependencyStatus::FinishedOrError); }
So foreach SalesParmTable found an instance of the runtime task FormLetterMultiThread is created, and is a dependency for the SalesFormLetterEndMultiThread to run.
Start by creating a RunBaseBatch class like you would otherwise do, but make sure that the code witch uses the most load is written in a separate method and called from the run. This method will be called from the threads. (method: updateSalesOrder)
The canMultiThread method is the same as in the FormLetter class.
protected boolean canMultiThread() {; return this.isInBatch(); }
And the run method could be written like this, analog to the run of the SalesFormLetter class, but without an ending thread.
void run() { int batchCounter = 0; ; try { ttsbegin; if(this.canMultiThread()) { batchHeader = BatchHeader::construct(this.parmCurrentBatch().BatchJobId); } while(this.queryRun().next()) { salesTable = this.queryRun().get(TableNum(SalesTable)); if(batchHeader) { tSTSalesOrderUpdateMultiThread = TSTSalesOrderUpdateMultiThread::newFromTSTSalesOrderUpdate(this); batchHeader.addRuntimeTask(tSTSalesOrderUpdateMultiThread,this.parmCurrentBatch().RecId); batchCounter++; } else { this.updateSalesOrder(); } } if(batchHeader) { batchHeader.save(); info(strfmt("%1 batches created",batchCounter)); } ttscommit; } catch(Exception::Error) { ttsabort; } catch(Exception::Deadlock) { retry; } }
The second class you need to create is kind of a wrapper class that also extends from RunBaseBatch and will be used to create the subtasks for your batch process. Make sure that the runsImpersonated method returns true.
Remember that you need to keep an instance of the caller class (TSTSalesOrderUpdate) and you need to pack and unpack it.
class TSTSalesOrderUpdateMultiThread extends RunBaseBatch { BatchHeader batchHeader; TSTSalesOrderUpdate salesOrderUpdate; container packedSalesOrderUpdate; #define.CurrentVersion(2) #LOCALMACRO.CurrentList packedSalesOrderUpdate #ENDMACRO } public static TSTSalesOrderUpdateMultiThread newFromTSTSalesOrderUpdate(TSTSalesOrderUpdate _caller) { TSTSalesOrderUpdateMultiThread instance; ; instance = TSTSalesOrderUpdateMultiThread::construct(); instance.parmSalesOrderUpdate(_caller); return instance; } public container pack() {; packedSalesOrderUpdate = salesOrderUpdate.pack(); return [#CurrentVersion,#CurrentList]; } public boolean unpack(container _packedClass) { int version = RunBase::getVersion(_packedClass); switch (version) { case #CurrentVersion: [version,#CurrentList] = _packedClass; salesOrderUpdate = TSTSalesOrderUpdate::construct(); salesOrderUpdate.unpack(packedSalesOrderUpdate); return true; default : return false; } return false; }
The run method should call the updateSalesOrder on your TSTSalesOrderUpdate class. This means that all the logic is placed in one place, because it should also work when not running in batch. 😉
void run() { ; try { ttsbegin; salesOrderUpdate.updateSalesOrder(); ttscommit; } catch(Exception::Error) { ttsabort; throw error("error"); } catch(Exception::Deadlock) { retry; } }
In addition you can add an ending multithread class if necessary, like the FormLetterEndMultiThread class. The maximum number of simultaneous batch thread can be defined on the SysServerConfig form.
Hi,
Here is a simple code snippet to create Batch jobs from code. This convenient when starting a heavy load job from a user interface and still keep the client responsive.
TSTSalesOrderUpdate tSTSalesOrderUpdate; BatchInfo batchInfo; BatchHeader batchHeader; ; tSTSalesOrderUpdate = TSTSalesOrderUpdate::construct(); batchInfo = tSTSalesOrderUpdate.batchInfo(); batchInfo.parmCaption("Test from code"); batchInfo.parmGroupId(""); batchHeader = BatchHeader::construct(); batchHeader.addTask(tSTSalesOrderUpdate); batchHeader.save();
Many projects use an interface to import their sales orders, because of this a SalesAutoCreate class was created. This class is easily extendable and customizable.The first thing to do is designing a buffer table, like this one for example:
After this we can start extending a new class from the SalesAutoCreate class and modifying the construct on the SalesAutoCreateClass .
class TSTSalesAutoCreate extends SalesAutoCreate { TSTSalesImport TSTSalesImport; }
static SalesAutoCreate construct(Common buffer = null, Object object = null, Common buffer2 = null) { switch (buffer.TableId) { case tablenum(SalesCreateReleaseOrderLineTmp) : return new SalesAutoCreate_ReleaseOrder(buffer,object,buffer2); case tablenum(PurchLine) : return new SalesAutoCreate_ProjPurchLine(buffer,object); case tablenum(SalesBasketLine) : return new SalesAutoCreate_Basket(buffer,object); // TST case tablenum(TSTSalesImport) : return new TSTSalesAutoCreate(buffer,object); // TST } throw error(strfmt("@SYS23419",tableid2name(buffer.TableId))); }
After the construct your buffer record should be passed in the new or a parm method could also be an option.
void new(Common _initFromBuffer, Object _callBackClass) {; super(_initFromBuffer,_callBackClass); TSTSalesImport = _initFromBuffer; }
When extending from the SalesAutoCreate class some methods must be implemented:
In my example I’ve implemented them like this:
protected void setCust() {; custTable = CustTable::find(TSTSalesImport.CustAccount); if(!custTable) { throw error("Custtable not found!"); } } protected SalesType salesType() {; return SalesType::Sales; } protected boolean recordExist() {; return TSTSalesImport.RecId != 0; } protected void nextRecord() {; next TSTSalesImport; } protected CustInvoiceAccount invoiceAccount() { CustInvoiceAccount ret = custTable.InvoiceAccount; ; if(!ret) { ret = custTable.AccountNum; } return ret; }
The next step is setting our table and line fields by overriding the setSalesTable and setSalesLine methods and make sure that you always call the super first. Notice that you need to call the createSalesTable and createSalesLine to do the insert.
protected void setSalesTable() {; super(); this.createSalesTable(); } protected void setSalesLine() {; super(); salesLine.ItemId = TSTSalesImport.ItemId; salesLine.itemIdChanged(); salesLine.SalesQty = TSTSalesImport.SalesOrderedQty; this.createSalesLine(); }
you should have a class looking like this.
the final step is to call the logic from a job or a RunBaseBatch class, make sure that you select records on the same tier as the nextRecord will run or else it will fail. Preferably on the server tier. 🙂
select forupdate TSTSalesImport; SalesAutoCreate = SalesAutoCreate::construct(TSTSalesImport); SalesAutoCreate.create();
This example lacks some validation whether the record has already been processed or not, so it will created the same records every time it is called. You could implement your own method on the SalesAutoCreate class, call it from the create method and override it on your custom class, like this.
void create() { #OCCRetryCount try { setprefix("@SYS55110"); ttsbegin; while (this.recordExist()) { this.setCust(); setprefix(#PreFixField(CustTable,AccountNum)); this.setSalesTable(); this.setSalesLine(); setprefix(#PreFixField(SalesLine,ItemId)); //-> TST this.deleteProcessed(); //<- TST this.nextRecord(); } this.endUpdate(); ttscommit; } catch (Exception::Deadlock) { retry; } catch (Exception::UpdateConflict) { if (appl.ttsLevel() == 0) { if (xSession::currentRetryCount() >= #RetryNum) { throw Exception::UpdateConflictNotRecovered; } else { retry; } } else { throw Exception::UpdateConflict; } } } //SalesAutoCreate protected void deleteProcessed() { } //TSTSalesImport protected void deleteProcessed() {; super(); TSTSalesImport.selectForUpdate(true); TSTSalesImport.delete(); }
While doing a small AIF project I wrote a small batch class to cleanup the AIF document log because the button on the AifDocumentHistory form can take up a huge amount of time. The first thing I did to write this class is checking out the standard Ax code in the following method ClassesAifMessageManagerclearAllProcessedAndError. This method uses a progress bar and deletes records in batches of 3000 records, this is something we don’t need when running in batch.
The first thing our method needs to do is check if we have access rights to delete.
if ( !hasTableAccess(tablenum(AifMessageLog), AccessType::Delete) || !hasTableAccess(tablenum(AifDocumentLog), AccessType::Delete) || !hasTableAccess(tablenum(AifCorrelation), AccessType::Delete)) { throw error("@SYS113226"); }
The second step is requesting the permission to skip AOS validation.
skipAOS = new SkipAOSValidationPermission(); skipAOS.assert();
The next step is calling all the skip methods, Microsoft does this to make sure that a delete_from doesn’t fall back to row by row deletes.
aifMessageLog.skipAosValidation(true); aifMessageLog.skipDatabaseLog(true); aifMessageLog.skipDataMethods(true); aifMessageLog.skipDeleteActions(true); aifMessageLog.skipDeleteMethod(true); aifMessageLog.skipEvents(true); aifDocumentLog.skipDatabaseLog(true); aifDocumentLog.skipDataMethods(true); aifDocumentLog.skipDeleteActions(true); aifDocumentLog.skipDeleteMethod(true); //BP Deviation Documented aifDocumentLog.skipAosValidation(true); aifDocumentLog.skipEvents(true); aifCorrelation.skipDatabaseLog(true); aifCorrelation.skipDataMethods(true); aifCorrelation.skipDeleteActions(true); aifCorrelation.skipDeleteMethod(true); //BP Deviation Documented aifCorrelation.skipAosValidation(true); aifCorrelation.skipEvents(true);
After these methods we can start deleting the records, I’ve used a utcDateTimeRemove variable to cleanup records after a certain number of days.
delete_from aifDocumentLog exists join aifMessageLog where aifDocumentLog.MessageId == aifMessageLog.MessageId && aifMessageLog.createdDateTime <= utcDateTimeRemove && (aifMessageLog.Status == AifMessageStatus::Processed || aifMessageLog.Status == AifMessageStatus::Error); delete_from aifCorrelation exists join aifMessageLog where aifCorrelation.MessageId == aifMessageLog.MessageId && aifMessageLog.createdDateTime <= utcDateTimeRemove && (aifMessageLog.Status == AifMessageStatus::Processed || aifMessageLog.Status == AifMessageStatus::Error); delete_from aifMessageLog where aifMessageLog.createdDateTime <= utcDateTimeRemove && (aifMessageLog.Status == AifMessageStatus::Processed || aifMessageLog.Status == AifMessageStatus::Error);
The final step is to revert the code access permission.
CodeAccessPermission::revertAssert();
Source : msdn xRecord class
(this job should never run on a production environment, build an archiving alternative instead)
(edit if you are using Ax 2012 have a look at this post Dynamics Ax 2012 History cleanup for a better solution)
Since i’m getting a lot of google hits on my Dynamics Ax – workdays to days post, i’ve decided to blog some more about it. The DateTimeUtil class is actually a wrapper of the .NET DateTime class.
A first thing to remember when using UtcDateTime EDT’s is that it is stored like the name says as Coordinated Universal Time.
The controls on the form will translate the DateTime to the timezone of the client. Now keeping this in mind is very important when mixing date, time and datetime controls. The following example will make it more clear.
The first field is a UtcDateTimeEdit control with a data method that returns DateTimeUtil::UtcNow().
As you can see the time is 08:58, but the first control on the form shows 10:58. This is correct because my client timezone is (GMT+01:00) Brussel, Kopenhagen, Madrid, Parijs and it’s summer time.
The second field is a TimeEdit control with a data method that returns DateTimeUtil::time(DateTimeUtil::utcNow()), this isn’t correct because it will always return the time in the UTC timezone and the control will not translate it to the correct timezone.
The third field is another TimeEdit control with a data method that returns TimeNow(), this is correct because the TimeNow method will also apply the client/server (depending on the tier) timezone.
this also applies to field in a table.
Another way to use Time controls and the DateTimeUtil is using the applyTimeZoneOffset method.
UtcDateTime ret; ; ret = DateTimeUtil::applyTimeZoneOffset(DateTimeUtil::utcNow(),TimeZone::GMTPLUS0100BRUSSELS_COPENHAGEN_MADRID); return DateTimeUtil::time(ret);
but this is a lot of code for a rather more simple thing 🙂
A second thing to remember, when querying with a date on UtcDateTime fields make sure you select the whole day from 00:00:00 to 23:59:59.
For this I like to implement a method on the Global class, it’s keeps you’re queries cleaner 😉 It works the same way as all .NET developers use, Add a with time 0 and subtract a second.
static UtcDateTime dateCompare(date _date, boolean _toDate = false) { UtcDateTime ret; ; if(_toDate) { ret = DateTimeUtil::addSeconds(DateTimeUtil::newDateTime(_date + 1,0),-1); } else { ret = DateTimeUtil::newDateTime(_date,0); } return ret; }
So you’re query looks like this.
Date currectDate = SystemDateGet(); SalesLine salesLine; ; while select salesLine where salesLine.createdDateTime >= dateCompare(currectDate) && salesLine.createdDateTime <= dateCompare(currectDate,true) { //Do something }
For testing purposes with the MRP we needed to modify the createdDateTime fields in Dynamics Ax 2009. Since these are system fields we needed a workaround.
if(isRunningOnServer() && CurUserId() == "Admin") { new OverwriteSystemfieldsPermission().assert(); salesLine.overwriteSystemfields(true); // "YYYY/MM/DD HH:MM:SS" salesLine.(fieldnum(SalesTable,CreatedDateTime)) = str2datetime( "2010/04/03 11:00:00" ,321 ); salesLine.doInsert(); salesLine.overwriteSystemfields(false); CodeAccessPermission::revertAssert(); }
Remarks:
Since this code is pretty exotic and you don’t want to release this to a production environment we eventually didn’t use this but ran some sql jobs, but this shows that it is possible.
edit
Easy does it 😉
DECLARE @DATAAREAID nvarchar(4) DECLARE @SALESID nvarchar(20) DECLARE @DATETIME datetime /* EDIT THE FIELDS BELOW */ SET @DATAAREAID = 'CEU' SET @SALESID = '00003352_058' SET @DATETIME = '01/01/98 00:00:00.000' /* DO NOT EDIT HERE */ UPDATE dbo.SALESTABLE SET CREATEDDATETIME = @DATETIME WHERE SALESID = @SALESID AND DATAAREAID = @DATAAREAID; UPDATE dbo.SALESLINE SET CREATEDDATETIME = @DATETIME WHERE SALESID = @SALESID AND DATAAREAID = @DATAAREAID;
Today I had to fix a bug in some custom code in Dynamics Ax 2009, we had an error from the JournalTransList class that stated “A key with the name %1 already exists.”
Ever had to calculate the number of days starting from a number of workdays and even deal with holidays.
A pretty straightforward example using the DateTimeUtil class and DayOfWk method.
static void WorkDaysToDays(Args _args) { int workDays = 10; int days = 0; date startDate = systemDateGet(); CalendarId calendarId = CompanyInfo::find().ShippingCalendarId; ; if(calendarId) { while(workDays > 0) { if(dayofwk(startDate) != 6 && dayofwk(startdate) != 7 && WorkCalendarDate::isDateOpen(calendarId,startDate)) { workDays--; days++; } else { days++; } startDate = DateTimeUtil::date(DateTimeUtil::addDays(DateTimeUtil::newDateTime(startDate,0),1)); } info(strfmt("Workdays = %1, Days = %2",workDays,days)); } else { error("No calendarId found"); } }