But after restarting the search services I received the error “search is offline” in the Search admin. An other symptom was that all crawls would keep on running. Even the people content source was running for more than 112 hours (after a weekend).

After some googling I found that I wasn’t the only one with this problem. Some resolved it by reconfiguring the search service account. Others deleted the ShareService provider and recreated it…

Fortunately I could resolve the issue by resetting the search index. After this everything worked fine again.

[sourcecode language="csharp"]
void main(void)
{
}
[/sourcecode]

And this will become:

void main(void)
{
}

Supported languages are:

• cpp

• csharp

• javascript

• powershell

• sql

• xml

• css

The complete list and help is at: http://en.support.wordpress.com/code/posting-source-code/

The Blog-Health-o-Meter™ reads Minty-Fresh™.

Crunchy numbers

A Boeing 747-400 passenger jet can hold 416 passengers. This blog was viewed about 3,000 times in 2010. That’s about 7 full 747s.

In 2010, there were 4 new posts, growing the total archive of this blog to 16 posts. There were 13 pictures uploaded, taking up a total of 1mb. That’s about a picture per month.

The busiest day of the year was October 26th with 29 views. The most popular post that day was Migrate SP2007 to SP2010, first impression.

Where did they come from?

The top referring sites in 2010 were google.co.in, google.com, bigextracash.com, statistics.bestproceed.com, and google.com.sg.

Some visitors came searching, mostly for spcontext.current.site, spcontext.current.web, spcontext.current.site.openweb, sharepoint in dotnet, and spcontext.current.web.lists.

Attractions in 2010

These are the posts and pages that got the most views in 2010.

1. Migrate SP2007 to SP2010, first impression January 2010

2. SPContext.Current.Web or SPContext.Current.Site.OpenWeb()? September 2009

3. Some tests with BizTalk 2006R2 WSS adapter April 2009

4. Move favorites to Dropbox February 2010

5. ProcessBatchData can be buggy February 2009

The query I used was:

After some Googling I suddenly realized that a few weeks ago I turned on indexing for some columns in the list. This was part of troubleshooting performance issues with this list, see also the previous post. Activating indexing for some columns didn’t resolve the performance issues (something else did) but I thought indexing is always good.

But it isn’t… After removing the specified column (Thema) from the index the CAML query worked fine again!

By the way: this isn’t only a problem for custom CAML queries. Also views that are configured to show only items with empty values won’t show the correct results.

Over the last few years the document library that this webpart was connected to was gradually filled with documents. Currently there are about 12000 items. This shouldn’t be a problem for SharePoint but downloading and zipping 20 documents of 100kb took about 3 minutes…

Diving in the code that was written some years ago I discovered that the SPList.Items was frequently used. After some Googling I found out that when the Items property of a list is accessed ALL items are retrieved from the database! Any additional function will be performed after the complete set is retrieved.

So when for example the following code is used:

all items are retrieved from the database and then the item with id 100 is returned. When this code is changed to:

Only the item with id 100 is returned.

Changing this resulted in a performance gain from 3 minutes to 1.8 second!

Cool…

So I decided to use the regedit preferred method but instead of SHGetKnownFolderPath I used the Set variant. Using Visual Studio I created the following c++ program. (sorry for the image, my code snipped plugin doesn’t work with the current wordpress theme)

This is a console application with one parameter which specified the new path. And this worked!

The compiled executable can be downloaded here: ChangeFavorites. It’s a zip file, so rename to .zip and extract!

There are two ways in which a SharePoint 2007 site can be migrated. One of them is database attach,which I used for this test. I used one of the first SharePoint websites we created so no fancy stuff with features and solutions, every deployment was done manually.

Steps I have performed:

1) Restore a SQL backup of the website in my SharePoint 2010 VPC.

2) Create a web application in SharePoint 2010

3) Attach the restored database to the web application with stsadm –o addcontentdb. Stsadm is a bit old fashion but it does the job. The database is around 600Mb and the stsadm command took about 10 minutes to complete. After 10 minutes it finished and pointed me to a logfile with some (lots) errors and warnings. The most common errors were about:
- features not found
- site template not found
- some custom webparts not found

4) After ignoring all errors I tried to give it a shot. I opened the website in a browser and it asked for logon credentials. But I don’t have any because the backup was from a SharePoint site in a different domain. In the old days I solved this by changing the primary or secondary site collection administrator in Central Admin. But where is this in SharePoint 2010? Couldn’t find it. Therefore I created a policy that my account has full control for the whole web application. That did the trick!

5) After logging on to the site a unclear error message was shown. So I edited the web.config to disable custom errors and show the friendly dotnet yellow screen error. Much better… Next I solved the errors step by step:
- put custom created dll’s in de bin of the web application
- put safecontrols in the web.config
- put AppSettings in the web.config

6) Finally I got some errors about files it couldn’t find in a 1043 folder… Right I forgot, I took the backup from a Dutch site. And there isn’t a Dutch language pack for SharePoint 2010 yet. So I copied the 1033 folder and renamed it to 1043.

Finished!

And the result looks pretty the same as in SharePoint 2007:

The design isn’t pixel perfect everywhere anymore. Perhaps the new core.css meshes this up a bit.

And for some reason the site actions menu doesn’t work.

Perhaps next time I try to implement the new SharePoint 2010 user interface and solve the site actions menu. And I will do some additional testing of the custom features of this site like BDC, custom search webparts, custom people search, Google Maps webpart, etc.

I think this is a nice result for a test that took just a couple of hours. I didn’t change any code (CSharp, masterpage, pagelayout, etc) it just worked!

But the Web object from the context has some weird behavior. For example adding a property to the property bag doesn’t always succeed:

SPContext.Current.Web.Properties.Add["MyKey"] = "MyValue";
SPContext.Current.Web.Properties.Update();
SPContext.Current.Web.Update(); // just to be sure ;-)

This code will execute successfully and for the lifecycle of the program the property is available but when a new instance tries to access the property it isn’t there anymore.

int count = SPContext.Current.Web.Lists["MyList"].Views["All Items"].ViewFields.Count;

This line of code will work on some sites but not on all sites… On some sites it will show the actual number of fields while other sites have a count of 2 and only the fields Attachments and Title.

Changing the code will always return the correct fields:

using (SPWeb spWeb = SPContext.Current.Site.OpenWeb(SPContext.Current.Web.ServerRelativeUrl))
{
SPList myList = spWeb.Lists["MyList"];
int count = myList.Views["All Items"].ViewFields.Count;
}

Strange… after three years of SharePoint programming I have never encountered this issue. And now I have solved two issues with it on the same day.

So in stead of using something like this:

string title = item["Title"].ToString();

It’s better to use this:

string title = item[SPBuiltInFieldId.Title].ToString();

This way your code is language independent. And if Microsoft for some reason decides to rename a field (probably not) your code still works.

There are currently only 4 hits when I Google for SPBuildInFieldId. Perhaps soon there will be 5…

I have installed BizTalk 2006 R2 and WSS 2.0 and WSS 3.0 side by side on one VPC.

The goal is to test the default behavior of the adapter. With the help of some custom orchestrations/pipelines/etc a lot more is possible.

WSS Adapter as send port

This is for sending messages from BizTalk into SharePoint.

Send a Word document (or binary file) to a document library

Configuration: FILE receive port with pass trough pipeline. WSS send port with pass trough pipeline and a filter on the receive port.

When a document is put in the receive location it is send to the document library in WSS 3.0. So far so good. Now I want to give it a nice filename instead of something like Msg-534d8936-34d3-40ac-8541-6b35457cbbb0. In BizTalk it is possible to use several macros and XPATH queries to generate a filename. But this is a Word document so XPATH queries won’t work here. So were bound to using macros. Out of the box there are the following macros: MessageID (GUID), SendingOrchestrationID (GUID), SendingOrchestrationType, Filename. The macros aren’t very helpful because they all end up with not human friendly names. The Filename macro creates by default something like Msg-534d8936-34d3-40ac-8541-6b35457cbbb0.

But the filename doesn’t have to be a real issue. SharePoint can automatically map document properties to list columns. If for example there is a document property Title and a document property Keywords and a list column Title and a list column Keywords then SharePoint fill them in with the correct values:

Or with custom properties like this Excel document (TotalCosts):

But with generated filenames with GUID’s it isn’t possible to overwrite (or update) documents:

Possible solutions are to create some additional BizTalk artifacts which set the WSS.Filename property. For example this can be done from an orchestration or a pipeline. Or this can be resolved by creating a SharePoint workflow which will be activated for each new document and rename the document to for example %title%.doc.

Create list item in custom list

Configuration: File receive port with XML pipeline. WSS send port with XML pipeline and a filter on the receive port and some values of the XML are mapped with the XPATH macro to the columns in the SharePoint custom list. Create and deploy a BizTalk assembly with the schema of the XML.

When an XML is put in the receive location a new list item is created in the custom list with the mapped values. The XML message itself is gone. A maximum of 16 columns can be mapped to the XML message.

It’s not possible to perform updates. A possible solution for updating items is described in the next test.

Create list item in document library

Configuration: File receive port with XML pipeline. WSS send port with XML pipeline and a filter on the receive port and some values of the XML are mapped with the XPATH macro to the columns in the SharePoint document library. Create and deploy a BizTalk assembly with the schema of the XML.

When an XML is put in the receive location a new list item is created in the document library with the mapped values. The XML message itself is also saved in the document library. A maximum of 16 columns can be mapped to the XML message. Because we are using xml we can do some nice stuff for generating filenames. For example the information inside the XML can be used. This way the filenames are predictable and BizTalk can overwrite list items.

And after an update (overwrite):

Document library settings

What is the influence of some document library settings which can potentially block BizTalk from creating or updating list items?

User has checked out a list item: BizTalk can’t update the list item. The message will be suspended (resumable) in BizTalk. After the user checks in the document the message can be resumed in BizTalk and the list item will be updated.

List setting -Require content approval for submitted items: BizTalk will create a new list item with the status pending.

List setting -Require documents to be checked out before they can be edited: Hmm… nothing happens! Apparently BizTalk can’t deliver the Message. But the message doesn’t appear in BizTalk as suspended. So the message is lost… Or not! BizTalk does send the document to the document library. But BizTalk hasn’t checked the document in. When I logon to the SharePoint site with the BizTalk service account the document is shown but checked out to the BizTalk service account.

List setting -Create major and minor (draft) versions: BizTalk creates a minor version. Or to be more specific two minor versions. The 1^st version only contains the message. The second version adds the metadata to the columns:

List setting -Create major versions: Same as the previous one. But now with major versions:

Now let’s update a list item and then check the version history:

The list item is updated. But there is no version history! Also interesting is that the title field is a generated name instead of the friendly name we have specified.

WSS Adapter as receive port

This is for reading messages from SharePoint.

Read from a document library

Configuration: WSS Receive port with pass through pipeline. FILE Send port with pass through pipeline and filter on receive port.

BizTalk polls the document library at a specified interval. When a document is uploaded to the document library BizTalk reads it from the document library and deletes it.

There are some configuration settings for the read port:

Archive: BizTalk always deletes the file from the document library after it is processed. But there is an option to archive the document. With these settings BizTalk can copy the document an archive document library.

View name: BizTalk can be configured to read all documents from the document library or only the documents in a specific view. In SharePoint a view can be created to show only documents based on some metadata properties. For example a view that shows only approved documents. If this view is configured then BizTalk will only process approved documents.

Read from a custom list

This is not supported: