Travails of an ungeeker

Dude, why can’t I add a Workflow or Form Library to Enterprise MOSS?

2009-07-05T12:15:00.001+02:00

Because in a publishing site some features are not activated by default. Activating the features below adds back your functionality you are used to on a collaboration site. I had this issue on two separate projects this week!

Ensure that the following features are enabled (on the site and the site collection):

Office SharePoint Server Enterprise Site features
Team Collaboration Lists
Office SharePoint Server Standard Site features

From: http://www.williamkent.net/archive/2008/06/09/configuring-sharepoint-for-use-with-infopath-and-forms-server.aspx

Visio Reverse Engineer error - cannot extract column definition

2006-12-18T02:51:00.000+01:00

Got this error because the tables were not owned by dbo. When I changed the ownership to dbo then the tables imported into Visio just fine. Tried importing (linking) the tables into Access using the same DSN and Access had no trouble. Surprisingly there were no posts about this when I Googled...

Copying a Visual Studio Setup (MSI) project

2006-11-10T04:54:00.000+01:00

In VS2003 setup projects are XML files, so you would think it is a simple matter if you are creating a bunch of related MSIs to copy one as a starting point and edit it. Well it does kind of work the steps are:

Open the .vdproj file and search for

"ProductCode" = "8:{189150D4-09F3-4998-B4B5-A2372E629E66}"
"PackageCode" = "8:{49B725DB-21AC-4FD1-AB84-95D5D2A9EA16}"
"UpgradeCode" = "8:{6394DA18-20B7-48CB-ACA6-C10B9DFC596B}"

Replace all three GUIDS by generating new ones - uppercase characters only! (select newid() in SQL returns GUIDS in uppercase).

Now when you save and open the copy in VS it works. You might get some strange errors, that will lead you to think this is like a corrupted binary file encoded in XML. Occasionally I have had to throw away a file that renders fine in VS but generates bad MSIs.

I'm still trying to decide if this approach is worth it myself, but it can save a lot of time.

Ouch my SQL Server collation orders are different and causing a collation conflict error

2006-09-01T00:29:00.000+02:00

You get the message: "Cannot resolve collation conflict " when you have multiple databases with different collation orders.

If one database was created in SQL 7.0 days, and took the default collation order you ended up with:
SQL_Latin1_General_Cp1_CI_AS

Then you created another database in SQL 2000, and took the default collation order you ended up with:
Latin1_General_CI_AS

Now you have a query, stored procedure or function call between these databases and you've got the conflict.

All the DB servers from DEV through to PROD should be set to the same database default to reduce this situation. If you've got a mixed up environment then you'll be coding around it forever, so bite the bullet and standardise your collation sequences everywhere.

There's a tool I haven't tried to fix this here:
http://www.codeproject.com/vb/net/ChangeCollation.asp

Problem with VS 2003 Setup Project default values in MSI

2006-08-25T02:54:00.000+02:00

Under User Interface, if you add a Textboxes dialog to collect a parameter for a folder path, and then supply a default path e.g. using Edit1Value then you might get this strange problem:

1) The user runs the MSI and edit the default say s:\wwwroot changing the value to c:\wwwroot.
2) Later when the user uninstalls the package from Add/Remove programs, the installer fails:

Invalid drive: s:\

Note that s: did not exist on install either. Interim solution: remove the default!
Anyone else have something like this?

ASP.NET validator controls throw an Exception in MCMS by default

2006-05-12T10:54:00.000+02:00

If you add an ASP.NET validator to an MCMS template you get a strange error like: MCMS Object has been deleted. An attempt was made to access the properties or methods of an object that was deleted by the current session.

First you'll find that a service pack for MCMS 2002 SP1a exists but which Microsoft discourages you from installing. Taking their advise and looking further you will find that the problem lies with the Web Author Console, where the standard processing of validator controls is not enabled by the default. Perhaps the service pack fixes that, but here is another way:

You can buy Stefan's book or just look at a posting from him here which shows the required code to add to make ASP.NET custom validation controls work.

http://www.gotdotnet.com/Community/UserSamples/Download.aspx
?SampleGuid=65F90778-465D-4A98-804F-2CCA72FF4704

There you will see you that you need to add some code to the AuthoringSaveNewAction method called by the Author Console. In your derived method, inject a call to Page_ClientValidate() which enables the validation support. That's for a new page save, so similar is requierd for AuthoringReeditSaveAction() and AuthoringReeditSaveAndExitAction().

Overriding the AuthoringSaveNewAction() of the console you need to inject a call to

if (this.Page.Validators.Count > 0) {
strReturn += "if (!Page_ClientValidate()) return false; ";
}

ScreenHunter 4.0 rave review

2006-05-10T18:12:00.000+02:00

These days I don't use many shareware utilities, but ScreenHunter is really a great screen capture / print screen replacement tool. It is especially useful for web-based user manuals because at the same as putting the image on the clipboard, it also creates a GIF (or JPG) file to use.

Other good features are: capture of the mouse pointer and tool tips (handy for marking a button of interest etc), and the rectangular copy, which is the main thing that Windows Print Scrn lacks.

Fix search problem in Articles module (for DotNetNuke) from Efficion Consulting

2006-05-10T17:43:00.000+02:00

First of all a prefix is warrented: this is a great module and does basically everything that our client wanted for news publishing, including the requirement for adding a separate archive page. It does this without having lots of complicated to set up options. Best of all, it's Free. Thanks Efficon.

However there is a problem with Search (the normal site search) if you do not use the categories feature in that none of your entries are ever matched during a search. At first it seemed that the ISearchable interface was not implemented (to tell DotNetNuke how to index the entries), however it is implemented.

I was very impressed with their support. We had some emails and they even logged onto our application server to have a look. However in the end the problem was only solvable by looking on the SQL Server, by running a SQL profiler trace, one call looked interesting:

exec dbo.Articles_GetArticles @PortalID = 25, @ModuleID = 3101, @Categories = NULL, @MaxNumber = 500, @Age = -1, @ShowAuthOnly = 1, @Featured = 0, @IgnorePublishDate = 0, @IgnoreExpired = 0, @SortField = 'CreatedDate'

This returned no rows. Trying this in SQL Analyer with @Categories = 0 did. Therefore you can add the T-SQL code below to the stored procedure Articles_GetArticles to implement a workaround without having to recompile the source. I am blogging this because despite initial helpfulness, I have had no response to my request that they confirm my fix (seems to be good) and implement this or a better one into their code.

if (@Categories is null)
begin
set @Categories = 0
end

ASP.NET User Control - accessing InnerHtml

2006-03-14T09:04:00.000+01:00

We have a user control using DHTML that displays a button which when clicked renders a floating DIV layer with an image. It is easy to define global string variables that are set from attributes of the user control to pass parameters for the image to show and caption, but then we wanted to inject HTML from between the opening and closing tags of the User control. Creating another element for this was ruled out because this would make the control too hard for our HTML designers to use.

I could not find a solution on the internet and until someone tells me a better one, here is a technique that works:

Define a global property in the User Control with public accessor called Span.
Place an opening and closing SPAN tag between the opening and closing tags of the User Control.
Place your HTML code inside the SPAN.

Now you can access your HTML with Span.InnerHtml. In the case we had we also did the following to inject the user's HTML into the HTML of the User control and did the following.

Add a Literal control to the design view of the User control in the place where the additional HTML should be injected.
In the User control code, override the Render method (Note constructor cannot be used as the control has not been instantiated yet)
Set the literal's Text property to the Span.InnerHtml property. (For gotchas, using Render, you have to check that your span object is not null because it is called several times, and remember to call base.Render(writer);

For the record, Blogger needs to provide some support for quoting code - especially indent formatting. I deleted the code I had from here because I could not make it render in Blogger acceptably. Sorry.

Timezone hassels in .Net - Daylight savings and change between UTC, GMT and others

2006-02-22T13:38:00.000+01:00

At least with .Net 1.1 it is suprisingly tricky to perform data entry in a timezone other than in machine time ("local time"), or in UTC (Universal Time = GMT - to which daylight saving is never applied).

It is interesting to read the comment below on gotdotnet about why Microsoft do not support this, even though it seems they clearly do that already internally for the regional settings on the control panel and all the data is there in the registry. For me to calculate CET (Central European Time) I cannot just add one hour to Universal Time because daylight saving needs to be considered. PS: According to Wikipedia it is Daylight Saving not Savings.

Now I'm looking at a CodeProject class that obtains the information from the Windows registry. Trouble is, that if the calculation is ever wrong, there will be the possibility of a press release for a major company going live an hour out, using Microsoft Content Mangement Server (MCMS). I got an email from them where they accept this risk.

Here's the GotDotNet quote:

Does the .NET Framework support Time Zone conversions to any given Time Zone?
Not in V1.0, V1.1 or the Whidbey pre-release.The .NET Framework does support conversion to and from UTC and the systems current local time. It can also support parsing a DateTime from an arbitrary time zone offset, such as 2003-10-26T13:11:07+10:00, but it must always convert this either to Local or UTC.This is a very common feature request and is likely to be in a future version. People are often surprised why this feature cannot be supplied by Microsoft at low cost. In particular, data to do conversions exists in the Windows registry and is used by the time zone selection dialog. However, there is a big distinction between having UI and registry data and having an API.This is a more expensive feature to undertake for Microsoft than most people would imagine because (a) an API must provide consistent behavior from one machine to another so we cant just re-expose the registry data and (b) there is cost for Microsoft in exposing an official Time Zone conversion because we face on-going geo-political costs for any country/region based data we gather and maintain. For example, a country may threaten to boycott our product if they are not listed in the data. This has happened to us with our CultureInfo data on many occasions, and we often need to tweak data in service packs, which is expensive and risky.That being said, there is agreement that this is a very important feature, and it is under serious consideration for the WinFx release.

source:

http://www.gotdotnet.com/team/clr/bcl/techarticles/techarticles/DateTimeFAQ/FAQ.aspx

I still had a problem where the calculation was not symmetrical and I was losing an hour. The Gotcha that hit me is that parsing dates in or out from text one needs to set Invariant culture

DateTime.Parse("d", DateTimeFormatInfo.InvariantInfo());

Compiler Error Message: CS1595 - LoadControl() issue when debug set to false

2006-02-06T09:17:00.000+01:00

A Google on this error returns two possibilities on microsoft.com (at time of writing). Neither documented solution applied here. However I remembered recently setting the web.config debug=false instead of true, which had triggered the error. I didn't have time to fully look into the problem, but it complained about a user control having two definitions, which really was true! The was being loaded dynamically with LoadControl() but there was still a static reference. I haven't thought through why it would have worked ok in debug mode. Anyone know? Perhaps the added debug hooks mask the error.

There could also be a bug somewhere in Visual Studio because the offending line was in the user control which in turn contained a script tag (client-side) with a src attribute. When the src attribute is commented out then the dynamic load of the user control caused no problems. In my case I can refactor this particular dynamic load out of the code anyway. Perhaps this points someone else similarly stuck in the right direction.

MCMS Site Manager Export 404 error

2006-01-26T22:59:00.000+01:00

Here's an undocumented solution:

Using Microsoft Content Managment Server 2002 SP1a, the Site Manager export would preview, but at the end of an export (during the write) stopped with the message: The remote server returned an error: (404) Not Found

The Microsoft FAQ entry entitled "Topic: 404 during Site Deployment" lists some tips that did not solve it for me. I just lost 3 hours on this and saw no mention on the web of the fix I needed on the web for this which follows:

The solution for me was to change the "Local Disk Cache Location" and set permissions on this folder. It was previously set to a to the MCMS server folder, which I saw evidence had been previously altered.

Ubuntu on VMware player is pretty cool too

2006-01-13T17:22:00.000+01:00

I've been playing a while with Ubuntu. I'm running it it under VMware Player and am very impressed. Having converted to Microsoft VirtualPC 2 years ago through MSDN and not trying it since - it is now clear the VMware are still ahead.

The Ubuntu installed very easily. I had the following initial fun:

default password for ubuntu is ubuntu - didn't see it documented anywhere but found it by trial and error.
Networking didn't initially work. Had to set the DNS server. Setting it didn't work at first, and then after it did, it reverted to an old IP address on me after a suspend/resume. Aftersetting it again it finally stays there.
Eclipse installed apparently fine using the built in "add package" command, but the configuration complains about missing files and will not let me install any additional packages (I want to add Mono).

I love Firefox

2005-12-20T09:34:00.000+01:00

I've been a fan since pre-release days, but it just keeps getting better and better. It's another Christmas (presents anyway) to discover wonderful new extensions. First and foremost are the developer extensions. Just in time, Microsoft are now offering something similar for IE. But it's difficult to see how Microsoft will be able to catch up with the success of FireFox extensions.

My current favourites:

JavaScript debugger (Venkman - currently requiring a fudge for 1.5)
LiveHttpHeader (see the trace of HTTP GETs etc in the side bar)
ProxySwitcher (currently not available for Firefox 1.5 ouch)
HTML validator (show validation summary of current page in status bar)
DOM Inspector (useful but needs work)
ColorZilla (allows you to sample a page to identify the colours used - beats using PhotoShop)
MeasureIt (allows you to measure distances in the browser - beats using PhotoShop)
ForecastFox (even though it insisted on snow all last week - for virtually nothing)

Just installed some new ones to try FireFTP, LinkChecker, TamperData, TestGen4Web

Dreamweaver MX File_Save.htm Javascript error in File_Save.js

2005-12-15T12:30:00.000+01:00

A colleague of mine was having this for 2 days. My suggestions were along the lines of Internet suggestions included reinstalling Dreamweaver and even reformatting! We didn't get to reformatting, but did get him to reinstall twice including a purge of files and registry keys missed by the uninstall first. No use.

Guess what? The problem was a corruption inside a file. Not a program file, but one of the files in a Dreamweaver project. Took too long to find because after all this was only a text file (.css). And we had already tried creating a whole new file and pasting the contents from the old one via a fresh copy of notepad. This copied whatever the corruption was. Doesn't make much sense, and how did the corruption get in into our file anyway. That's the kind of problems I used to get 10 years ago, seldom now.

Moral of the story: next time this Dreamweaver error happens, don't reinstall. Create a clean new file in the project with no content and rebuild the file. Don't paste the old text in.

DotNetNuke 3.2/4 HTML validation failures

2005-12-01T11:15:00.000+01:00

I am using DotNetNuke for more and more customer CMSs - instead of using big commercial products. However one area that really needs "tidying" up is the HTML quality. Now that the Microsoft ASP.NET team have addressed this I am really looking forwards to the DotNetNuke team doing likewise. Lets hope for full pages served with XHTML compliance, valid HTML and accessibility issues improved. Add TABLE-less design. I reluctantly added a table to a design of my just-finished skin because I could not get the menu to "live" inside a DIV. All the modules should be subjected to an HTML review.

This is really important if more big clients are to move to DotNetNuke. They care about these things.

DotNetNuke "Redirection limit for this URL exceeded"

2005-11-30T15:31:00.000+01:00

Got this one yesterday when reconfiguring some portals. Solution was that I needed to create the sub directory reflecting the child portal alias I had created. This was not automatically done by DotNetNuke when adding an alias. Thus my alias was www.company.com/iberia and for http://www.company.com/iberia to operate, there should be a folder named iberia with a default.aspx page that performs a redirection into the portal system.

DotNetNuke mjj.PageElements module

2005-11-23T15:22:00.000+01:00

This is a really handy extension to DotNetNuke which is offered free on the developers web site, including source. It lists all the sub pages as links, and you completely control the Html generated through a simple template system that requires no learning.

Unfortunately there is a bug where deleted pages are still listed. Without studying the code - most likely it needs to add "where IsDeleted=0" to the query that gets the subpages to list.

The workaround, if you don't want to get involved in maintaining code from other developers like me, is to go to the recycle bin and permanently delete the page.

Google can produce bad Url references (href) on a cached page by injecting an incorrect BASE tag

2005-11-22T11:01:00.000+01:00

A large client site is generating a lot of invalid page reference errors when the referrer was the Google page crawler engine (Googlebot). Google injects a BASE tag into the cached version of the page. In our case, because of a redirect to a language subfolder, this was wrong. If you used Google to find the site, but clicked on the cached version you could see that relative Urls were wrong.

Now the site has it's own base tag - set to the correct subfolder. Lets see if Google will leave our base tag alone next time the page is regenerated. Hopefully it doesn't overwrite the tag with a bad path as before.

PS: sorry I didn't manage to ungeek this issue at all!

HTML quirk of the day: Italics bug with Internet Explorer

2005-11-15T17:25:00.000+01:00

Only on some PCs and only on Windows, there is a lovely bug with italics blocks rendering too wide. On one customer site this forced the content area a few pixels, which was enough to move a whole left hand side menu down to the bottom of the page. But only on some machines. Both my colleague and I have the same IE (from Windows XPSP2). I get the problem he does not. Specifying italics in CSS instead of an HTML tag doesn't help either.

In our case forcing some line breaks was an acceptable workaround to keep the text from getting too close to the margin it was forcing too wide.

http://www.positioniseverything.net/explorer/italicbug-ie.html

You've gotta love web development.

Release of DotNetNuke 3.2 / 4.0

2005-11-08T10:28:00.000+01:00

This flurry won't continue, but I am very excited to see that a new version of DotNetNuke was just released. I have to have a hunt to see if some of the anonymous bugs that I logged have been fixed. So far only the new features are promoted.

IIS 6.0 Authentication stopped working

2005-11-07T10:02:00.000+01:00

Today a (virtual) machine of mine would not authenticate to an ASPX page. I would only get error 403.1 Finally found that Windows server 2003 Service pack 1 changes the integrated authentication formerly called NTLM in favor of Kerberos. I changed the joined of this new VM to a new domain based on Active Directory without created a clean build of the machine and got these strange problems. Finally found a hack to my problem in an article on Technet "Forcing NTLM Authentication (IIS 6.0)".

Note that editing the metabase in notepad to make the setting didn't work the first time. Not sure if this was because I had left the file open or entered it wrongly, or a restart was required (recall reading that a restart is not necessary).

This blog is a personal record of lessons learned - for when they happen again. I think only problems like this one that were not solved within a few google searches will get logged. Lets see...