Monday, 10 November 2008

Hi all

So, I finally decided to try out the FarPoint BizTalk adapter for Excel spreadsheets. It's always nice to have tried as many adapters as possible, so I can use this knowledge when talking to customers.

This post is the first of a series of posts about this product. The first post is about installation and the basic functionality. The next posts will go deeper into separate functionality.

Let me just make one thing clear before I begin: I am not in any way affiliated with FarPoint Technologies, nor are they paying me anything to write these posts.

So, to begin; The installation was easy. Just a next-next-finish wizard. All the information needed from you is the product code, a serial number, and the installation folder, if you want to change that.

After installation, I have the functionality I would expect from this kind of product:

  1. Pipeline components to use in pipelines
    excel_disassembler_component excel_assembler_component
  2. Schema extension for the schemas
    image
  3. Wizard to help me create the schemas
    add_schema
  4. Documentation
    spread_documentation

So, to sum up:

  1. The installation is easy.
  2. The architecture of the solution seems to be exactly what I would expect, ie. pipeline components, schema extensions and a wizard.

My next post in the series will take a deeper look at the wizard for creating schemas for use by the pipeline disassembler and assembler.

--
eliasen

Monday, 10 November 2008 23:49:25 (Romance Standard Time, UTC+01:00)  #    Comments [2]  | 

Hi all

In almost all multiple server installations of BizTalk I have encountered, there has been issues with MSDTC. MSDTC is Microsofts product for handling distributed transactions, meaning transactions that span multiple servers. BizTalk uses this in high scale, when running transactions against SQL Server, to maintain consistency in BizTalks databases.

All issues with MSDTC are solvable - sometimes it is just hard to figure out what is wrong.

First of all, always use the DTCTester tool at http://support.microsoft.com/kb/293799 to test your MSDTC installation. If this tool reports no errors and you are still having issues, then most likely, MSDTC isn't the cause of your issues.

If something is wrong with MSDTC, I have encountered four major issues:

  1. MSDTC doesn't run on either of the server. Solve this by starting MSDTC. Steps to start MSDTC (Note, that the MMC snapin is buggy, and it might appear that the "Component Services" node has no children... but it does, trust me :-) ):
    1. Go to "Administrative Tools" => "Component Services"
    2. Go to "Component Services" => "Computers" => "My Computer"
    3. Right click "My Computer" and choose "Start MS DTC".
      start_msdtc
  2. MSDTC isn't configured for network access on both servers. Solve this in "add/remove windows components" here:
    install_msdtc_network_access
  3. The two servers have the same MS DTC ID. This ocurs if both servers are clones of the same server or if one of the ervers is a clone of the other server. Usually, when cloning servers, sysprep is used to clear out those errors, but in case it hasn't been used, here is how you fix it:
    1. Run "msdtc -uninstall" from a command prompt
    2. reboot
    3. Run "msdtc -install" from a command prompt
    4. reboot
  4. You can't ping the servers by hostname, which is required. This basically means, that from both servers, you need to be able to ping the other server by hostname - pinging by IP address isn't enough. If you can't ping by hostname, you have two options:
    1. Get the network administrator to update your DNS
    2. Enter new information into the hosts file in c:\windows\system32\drivers\etc

Hope this helps.

--
eliasen

Monday, 10 November 2008 19:28:38 (Romance Standard Time, UTC+01:00)  #    Comments [0]  | 
Sunday, 09 November 2008

Hi all

Finally, I have managed to end the upgrading-hell it has been. I am now up to version 2.2.8279.16125 of dasBlog, and this is currently the latest release. I didn't have the guts to upgrade directly from 1.9 to 2.2, so I went over 2.0 and 2.1 to get there.

To me, everything looks ok - let me know if you discover anything else.

--
eliasen

Sunday, 09 November 2008 02:28:27 (Romance Standard Time, UTC+01:00)  #    Comments [1]  | 

Hi all

I have just upgraded dasBlog to 2.1.8102.813.

I hope this hasn't introduced weird behavior. Please contact me if you encounter errors or issues.

Thanks

--
eliasen

Sunday, 09 November 2008 01:15:46 (Romance Standard Time, UTC+01:00)  #    Comments [0]  | 

Hi all

I have just upgraded to dasBlog version 2.0.7226.0.

I am hoping it went without issues, but would appreciate an email if you discover any issues.

Thanks.

--
eliasen

Sunday, 09 November 2008 00:47:38 (Romance Standard Time, UTC+01:00)  #    Comments [0]  | 
Thursday, 23 October 2008

Hi all

The CSD/BizTalk Product Group is conducting a survey to assess customer usage of the Business Rules Engine (BRE).

They have sent a link to us MVP's to get our opinion, but would also like the opinion of BizTalk consultants, BizTalk customers, and so on.

If you can spare a few minutes, please take the survey at https://live.datstat.com/MSCSD-Collector/Survey.ashx?Name=BRE_Usage_Survey_Blog

--
eliasen

Thursday, 23 October 2008 20:57:55 (Romance Daylight Time, UTC+02:00)  #    Comments [0]  | 
Tuesday, 14 October 2008

Hi all

Today, a former customer of mine called me and had a question. She is using updategrams to insert rows into a table in SQL Server, and she had a problem with dates. The input CSV sometimes had an empty string for a date.

This would translate in the XML to an element with an empty string inside it. This would then get mapped to an attribute for that column in the table, and when SQL Server read the updategram, the 01-01-1900 was inserted into the date field because no other value was specified.

The customer wants null to be inserted instead. She asked me how to do this, and I said that that was easy - I would send her an example later this night.

So I started preparing my example, and it turns out, that my approach only works for elements and records - I have never before had to suppress an attribute, so I wasn't aware of this. My approach being, bye the way, that I connect a logical functoid to the destination node that I don't want created under some circumstances. But a logical functoid can not be connected to an attribute in the destination schema. Bummer.

So I eventually went with a custom scripting functoid, and my example ended up like this:

Input schema:

image

Just a plain schema with three elements, all required.

The output schema:

image

This schema has three attributes to simulate the updategram and also an element just to show the approach for elements.

The map ended up like this:

image

Element1 goes to att1 and Element3 goes to att3. No magic there.

As for element2 in the output, I have used to functoids to get it to be created only when a string is present in the source. The first functoid (the red one) is a "String Size" functoid, which returns the length of a string. The green functoid is a "Not equal to" functoid, which has the string length as one parameter and the constant "0" as the other parameter. The order of these parameters do not matter. So basically, at runtime, the element2 element is created if and only if, the length of the string in Element2 is larger than 0. For this particular customer, I could have used the "Logical Date" functoid instead of the combination of these two functoids, since that will return true if the input string is a date, which is exactly what I want. But the above solution is more general and applies to a lot more scenarios. Downside being, of course, that the above solution will accept a "abc" as a string, and since the length is larger than 0, it will get to the output.

Anyway, the customer didn't have elements, but attributes. for this I had to use a custom scripting functoid with this code:

image

Basically, this call template takes one parameter, being the element in the source that might be empty. It then check the value to see if it is different from the empty string, and if it is not the empty string, an attribute is created and the value for the attribute is set to the input parameter.

 

Now, as a side node, the input my customer has is a CSV file, so actually, in this case you can also take advantage of a nice little property on the schema. To demonstrate, I have created a flat file schema for a CSV input:

image

As you can see in the properties window, I have set "Suppress Empty Nodes" to "Yes". When BizTalk parses the flat file, the nodes that would be created as empty nodes now aren't created at all. This means, that the map can just look like this:

image

No functoids required, and it will give the exact same output as the above map. Do notice, that this only works with flat files where you make BizTalk suppress empty nodes.

So, that was all... feel free to ask questions :-)

You can find my demonstration project here

I hope this turns out to be helpful to someone.

--
eliasen

Tuesday, 14 October 2008 03:11:24 (Romance Daylight Time, UTC+02:00)  #    Comments [0]  | 
Tuesday, 30 September 2008

Hi all

Today i read an email from a guy who had a problem with optional elements in an input giving problems in a positional flat file output. The issue being, of course, that if an element in the input of a map is optional, it might be missing. If the element is missing, it will not be created in the destination of a map, and therefore, the flat file assembler will complain because it needs the element to create the correct positional structure.

I seem to have it working, and will here walk through my solution to explain it.

First of all, I have a Schema for the XML input:

All elements are 1..1 except Element2, which has minOccurs=0.

Secondly, I have a schema for the flat file output:

All elements are mandatory. The record delimiter is 0x0d 0x0a and the two subrecords to the root are positional records.

The map is pretty straight forward:

Just plain connections between the elements.

For testing purposes, I have two test instances, that validate against the input schema. They are exactly the same, except one doesn't have the "Element2"-element in it.

If I try to test the map with the input that has the "Element2"-element, and turn on validation of both input and output, and let the output be "Native", then it will work. If, however, I test the map inside Visual Studio .NET with the example that does not have the Element2 element, it will fail. It will report that:

Output validation error: The element 'FirstElements' has incomplete content. List of possible elements expected: 'Element6'.

So basically, the map does not create the Element6 element in the destination schema, and since the Element6 element is required, it fails validation.

BUT, here comes the surprise; It works if it is deployed. So basically, there must be some inconstency between how the map tester in VS.NET works and how the stuff works when running.

 

I tried changing the schemas to include an element inside the first record as the last element, such that the input has a "SomeElements" record with three elements inside it, of which only the second is optional. Likewise I added a new element in the output schema and updated the map. You can see all three here:

 

Still, I only get errors when testing inside Visual Studio .NET and not when things are deployed and running... Which actually bugs me a bit, but that is a whole other story.

So, to sum up, I only have three explanations as to why it works for me and not for the fellow with the issue:

  1. He is using a BizTalk version that is not BizTalk 2006 R2
  2. He hasn't tried deploying it, and is relying on the map tester
  3. He has some bogus values for the two properties I will mention below

At the end of this post, let me just quickly mention to properties that are available for flat file schemas:

The "Generate Empty Nodes" and "Suppress Empty Nodes" properties might be helpful. They are defined here: http://msdn.microsoft.com/en-us/library/aa559329.aspx

Hope this helps someone.

You can find my project here: FlatFileEmptyElements.zip (27.55 KB)

--
eliasen

Tuesday, 30 September 2008 01:01:20 (Romance Daylight Time, UTC+02:00)  #    Comments [3]  | 
Friday, 05 September 2008

Hi all

I have just been informed by my MVP Lead, that Patrick Tisseghem, a very excellent SharePoint guru and co-founder of U2U has passed away a couple of days ago.

I have had the pleasure of listening t Patrick talk only once, but he left an impression of a very knowledgeable and enthusiastic professional, as well as a great guy who is happy to meet people and talk to them about what ever is one his/their minds.

My thoughts go to his wife, his children and the rest of his family.

--
eliasen

Friday, 05 September 2008 21:04:06 (Romance Daylight Time, UTC+02:00)  #    Comments [0]  | 
Tuesday, 02 September 2008

Hi all

I am probably the 234'th blogger to write about this, but just in case...

Alan Smith has decided to maintain two bloggers guides for now... One is the commonly known Bloggers Guide to BizTalk. The new one is the Bloggers Guide to Oslo, which deals with the oslo related technologies, such as WCF and WF.

Alan has had the guide on gotdotnet earlier, but now he has created the http://bloggersguides.net/ site, which I highly recommend you monitor closely :-)

That's all, folks.

--
eliasen

Tuesday, 02 September 2008 22:44:04 (Romance Daylight Time, UTC+02:00)  #    Comments [0]  | 

Theme design by Jelle Druyts