Agile costing

A friend of mine +Nicholas Vanderpyle and ex-coworker was asking questions about agile costing. I thought I'd give it a try to answer him:

Question:
Anyone do agile costing in their software projects? I was in a meeting that segued into a discussion about how to cost and whether to adjust the cost or create a new ticket when new effort is discovered.
I'd love to pick your brain offline about how you do agile planning and costing. It'd be nice to see how someone else does this.

This is my "basics" answer. They are probably a lot more variables to take into account, but this is what comes to mind right away:

Simplify costing.

* Use the fibonacci sequence. 1, 2, 3, 5, 8. Anything greater than 8 is probably too big to be a single story. Break it down to 2 or more.

* Only cost things that add business value. Bugs do not add business value, they should have never happened in the first place, that's why we use testing (unit/automated/regression)

* "Research stories" are that, research. Do not cost them, they do not add business value. They will add value once we implement them. Even better, research stories should not be "stories" but "chores" or "internal dev tasks" or whatever name suits your process.

* To avoid never-ending discussions when costing stories, assign someone as the "master of ceremonies." That person's job will be drive the session, moderate discussions, set a time limit for the session (or discussion), and have the final word on costing if no consensus is reached.

That's all I can think of at the moment.

Few recommended readings:
* How to Estimate effort
* A case for costing bugs
* A case for not costing bugs
* How to split user stories

Hope this helps!

-D

Looking for a good bug tracking system? Give Countersoft Gemini a try.

Doing software trials when hunting for a bug tracking system can be a really painful experience(or for any kind of software solution for that matter). Between choosing your top 3 (or more) candidates and going through the comparisons of features, performance, cost, compatibility, extensibility, etc..., you will probably get to the weekends with no joy for life :) (Not to mention having to convince your co-workers that this is the way to go).

So Hopefully, I can point you to the right direction and save you a bunch of research time. I have been using Countersoft Gemini bug tracking software for about 6 years now. Actually, it has now evolved now into a pretty full featured Project Management Software without loosing much of the simplicity I was originally drawn to. I just want to point out that I have not blindly being using Gemini for the sake of it; As I've switched jobs and companies, I've had chances to evaluate and use other bug tracking and project management systems for short periods of time. However, in the end, Gemini always turned out to be the better system, and it was not difficult for my co-workers to see why.

So here are my Top 3 reasons of why I have stuck with Gemini over the years and why I believe it is a must-have software tool for any development team.

1. Performance and customization: 

   There is nothing more frustrating to me than a sluggish piece of software. Gemini is fast.  After the initial setup and customization, the most common task that one will do is search. Whether you are viewing issues from a specific version number, or the issues assigned to a resource, or issues that belong to a component, all is happening is a search with appropriate filters.  As projects grow, one would expect a bit of performance issues on lookups.  However, I don’t know what kind of optimizations these guys do on the backend, but lookups are fast.
   Another great feature they’ve got is customization.  Although, you can use everything right out of the box, gemini lets you customize pretty much anything:  Issue Types, statuses, severities, workflows, etc...

2. Add-ons 

   Everyone uses systems differently.  Some people like to use desktop apps, others rather use web interfaces, and a few like to have everything integrated in one place.  The Gemini team did their homework and has managed to provide different interfaces to choose from.  You can go the standard way and use the web interface, or you can download Gemini Desktop, which not only works just like the interface, but it also provides an integrated way to do screen captures and issue time tracking.  However, my favorite add-on is the Visual Studio 2010 add-on:  You can use Gemini without leaving Visual Studio, work through your “work queue”, update items, comments, do screen captures and manage the time spent on specific tasks.  The best part of this add-on is that it opens tickets as new tabs, so you can use the entire surface or you can dock it in an more subtle window. Granted, it is not as fast as the website because it goes through its own web services API, however, its perfect for managing and working individual tickets.  Which brings me to another quick point: If you want to integrate gemini into your existing intranet website or winforms app, etc… You can! They have a pretty comprehensive REST based web services API that lets you pretty much do anything you need to manage issues. Lastly, they have other great add-ons and products like the MS Outlook add-on, to view tickets as tasks, the Gemini scheduler, to create tickets from a pop3 mailboxes, the twitter add-on, source control repo connectors, Gemini iPhone App, etc…  Check them out, it’s impressive.

3. Support

   I think it’s very common for companies that have a really good or at least a decent product, to forget about how important supporting their customers really is.  When documentation is not enough, you’ve got to ask questions and its very frustrating to struggle to find answers.  This is where I was very pleasantly surprised by Gemini’s Support Forum.  Most of the questions are answered by the lead developers like Saar Cohen or Mark Wing. The really impressive part of this is the quick turn around:  I’ve had questions answered in less than 20 minutes in multiple occasions.  One of my co-workers had a big email thread going with these guys about how to solve an issue.  These guys are knowledgeable (of course, they built the system) and super helpful.  I can appreciate when developers go the extra mile to ensure their customers are taken care of and their products works like they should.  That alone is worth gold to me.

Hope this helps!

-D

ASP.Net 4 and IIS6

A few days ago I posted about how to switch frameworks (to ASP.Net 4) on a single website  (among many) without having to restart IIS.  Well, Now I ran into another issue:  “Connection Reset” error from the browser. Tried a bunch of things, including checking logs, recreating the App Pool, rechecking all website settings, etc….  Nothing. 

After some googling, it turns out that it is not enough to Install asp.net 4 but it has to be manually “enabled”.  I’ll give you the quick run through but here is the post that that goes into more detail about the fix.

1) Confirm that ASP.Net 4.0 is disabled.

cscript iisExt.vbs /ListFile

If the entry for .Net 4 has a 0 on the left, then you need to enable it

2) Enable .Net 4 extension

cscript iisExt.vbs /EnFile C:\WINDOWS\Microsoft.NET\Framework64\v4.0.30319\aspnet_isapi.dll

Run the first command and you should have a 1 instead of 0 next to .Net 4.

That did it for me, I hope it works for you as well.

-D

Using IIS Virtual Path: Upgrade a site to .Net 4 without restarting IIS

Recently, our project was upgraded to ASP .NET 4.0. Upgrading Dev environments was a breeze, however, in production, only one website needed to run ASP.NET 4.0 while the other websites needed to be left alone.

First thing I tried doing was go to the properties dialog of the website and switch the targeted framework.

This is the warning message I got:

Naturally I cancelled that and attempted running the suggested command, but the problem was how to find out the "IIS Virtual Path" bit?.

I found this answer in this Stack Overflow: http://stackoverflow.com/questions/1804208/how-do-you-work-out-the-iis-virtual-path-for-an-application

The answer basically boils down to executing "aspnet_regiis -lk" command to find out your website ID and then run the command suggested by the warning message.

My problem running "aspnet_regiis -lk" was that I got an incomplete list of IDs and also I didn't know which ID corresponded to the Website I wanted to work on.

A co-worker pointed out an easier way to find the IDs for your websites: click on the "Website" node (folder) in IIS. Now on the right side you should see a list of all websites with their "Identifiers", State, IPs and ports.

Cool! Now I can run my command with the correct ID.

1) Navigate to C:\Windows\Microsoft.net\Framework64\v4.0.30319
2) Execute aspnet_regiis.exe -norestart -s “W3SVC/YOURWEBSITEID/root”

Note: If your application is not at the root of the website, then you need to add the AppName at the end: “W3SVC/YOURWEBSITEID/root/AppName”

Now, go to your website properties and check that your targeted framework is ASP.NET 4.0 and voila!

-D

VS2010 New feature Call Hierarchy and ReSharper's Call Tracking

I've been playing with VS2010 for a couple of days now, and I just recently stumbled upon an awesome features that every Dev will love: "Call Hierarchy".

In simple words, you can now view where your methods/properties are being called from. Now, you might say, "We've had that for a while, its called 'Find usages'", Well, this is better: You can actually view and go through the call tree graph WITHOUT having to be in debug mode. As Borat would say: "Itzanice!"  Scott Gu has a really nice post about it, go see it.

Playing around with it, I noticed that ReSharper also has its own implementation of it: “Call Tracking”.  And, it is actually better  than the one with VS2010.  So, if you have ReSharper, go to a method and do Ctrl+Shit+Alt+A , then select “Incoming Calls”. A new window will open up showing all callers to this functions and the tree graph. It’s so cool!

 Select “Incoming Calls”

New Windows opens up with all callers and the call tree

You might have also noticed that you can select “Outgoing Calls” which is the exact opposite of viewing all the callers, it’s a “Go to definition” on steroids!

As ReSharper puts it, Call Tracking is “an all-out Find Usages and Go To Declaration”.  Go check out their post about it to find out more details.

Upgrading your build server to use .Net 4 + Nant

At work we use CruiseControl.net + Nant for our continuous integration. Recently we have upgrades to VS2010 and .Net 4.0 (awesome so far). Alas, as soon as we converted our solution to .Net 4.0, our build server broke! After googling and wrestling for 2 days with build errors, I finally got the build server running again. I found most of my questions on Stack overflow, however I had this "AL exited with code 128" build error that I could not get past and it was driving me crazy! So, below are the steps to upgrade to .Net 4.0 and to solve the "AL" error message (that hopefully you won't get):

First thing. You DO NOT need to intall VS2010 in the build server like many people think. You do have to install the latest Windows SDK. (I believe 7.0)

1) Install .Net 4.0 Framework.

2) Install latest Windows SDK.

Install the latest Windows SDK, then go to the registry and copy the key "[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0" and rename all instances of 7.0 to 7.0A. The Build server is going to look at that path in order to get the "Installation Folder key". Alternatively, You can copy the text below into a notepad and save it as "whatever.reg", then run it and it should import the key and subfolders into the registry.

*~~~~~~~~~~~~~~~~~~~~ Copy everything below this line ~~~~~~~~~~~~~~~~~~~~~~

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A]
"InstallationFolder"="C:\\Program Files\\Microsoft SDKs\\Windows\\v7.0A\\"
"ProductVersion"="7.0.30319"
"ProductName"="Microsoft Windows SDK for Visual Studio 2010"

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A\WinSDK-NetFx35Tools]
"ProductVersion"="7.0.30319"
"ComponentName"="Windows SDK .NET Framework 3.5 Multi-targeting Utilities"
"InstallationFolder"="C:\\Program Files\\Microsoft SDKs\\Windows\\v7.0A\\Bin\\"

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A\WinSDK-NetFx40Tools]
"ProductVersion"="7.0.30319"
"ComponentName"="Windows SDK Tools for .NET Framework 4.0"
"InstallationFolder"="C:\\Program Files\\Microsoft SDKs\\Windows\\v7.0A\\Bin\\NETFX 4.0 Tools\\"

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A\WinSDK-SDKTools]
"InstallationFolder"="C:\\Program Files\\Microsoft SDKs\\Windows\\v7.0A\\Bin\\"
"ProductVersion"="7.0.30319"
"ComponentName"="Windows Common Utilities"

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A\WinSDK-Win32Tools]
"InstallationFolder"="C:\\Program Files\\Microsoft SDKs\\Windows\\v7.0A\\Bin\\"
"ProductVersion"="7.0.30319"
"ComponentName"="Windows Utilities for Win32 Development"

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A\WinSDK-WindowsHeadersLibs]
"InstallationFolder"="C:\\Program Files\\Microsoft SDKs\\Windows\\v7.0A\\"
"ProductVersion"="7.0.30319"
"ComponentName"="Windows Headers and Libraries"

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SDKs\Windows\v7.0A\WinSDK-WinSDKIntellisenseRefAssys]
"InstallationFolder"="C:\\Program Files\\Reference Assemblies\\"
"ProductVersion"="7.0.30319"
"ComponentName"="Windows Intellisense and Reference Assemblies"

*~~~~~~~~~~~~~~~~~Do not include this line~~~~~~~~~~~~~~~~~~~~~

3) Modify your Nant.exe.config file.

Towards the end of your Nant.exe.config file, modify the <startup> section to add .Net 4.0 as a supported Runtime:

<supportedRuntime version="v4.0.30319" />.

Also, under the <properties> section, add a property:

<property name="nant.settings.currentframework" value="net-4.0" />

This will make your builds default to that framework. You can also put this property in your projects Nant file, so you can tell each project what version to target. Finally, you need to add the .net4.0 framework node. Here is a link to that configuration section: http://pastebin.com/3ghS0gyA

4) Review the Build section of your Nant file

I was using the msbuild task for NantContrib. That still works, but if it doesn't, you can always use the <exec> task and call msbuild.exe like so:

<property name="msbuild.exe" value="C:\WINDOWS\Microsoft.NET\Framework\v4.0\MSBuild.exe" />    

<exec program="${msbuild.exe}">        
        <arg line="/property:Configuration=${YourSolutionConfigurationHere}" />         
        <arg line='"${YourSolutionFileFullPathHere}"' />         
        <arg value="/target:Rebuild" />         
        <arg value="/verbosity:normal" />         
        <arg value="/nologo" />         
</exec>

5) Random Errors

I ran into some weird build errors: "error MSB6006: "AL.exe" exited with code 128" That's it, not helpful at all. I did some googling and came up with few answers. Here are the links: msdn help and asp.net forums.  I wrestled with that error for a full day, studying the "detailed" and "diagnostic" logs, but all it did is pointing me to the assembly that failed building. No specific error. I could not even duplicate it on my local box.

Finally, I decided to try the suggestion about the "resource" files naming convention in the asp.net forum link, and... alleluia! my build started working. I don't know what's up with the build failing because of the resource name, still working on that, but my immediate goal was to get the build working.

StackOverflow related posts:

http://stackoverflow.com/questions/1215731/building-net-4-projects-with-nant
http://stackoverflow.com/questions/2353404/nant-msbuild-task-with-net-4-0-rc
http://stackoverflow.com/questions/1291221/going-from-solution-to-exec-programmsbuild-in-nant

Hope this helps someone else out there.

-D

Range() and LinqToXml

A couple of days back, I was working on some dashboard components for our portal that use fusion charts. I was generating empty XML templates when I ran into a bit of a challenge: I had to generate 30 "category" nodes that were, basically, dates, and also empty data values for each of those dates. The trick was that the dates had to be based on the day the Chart was accessed.

Well, to be fair, it wasn't really "a challenge", just one of those pain in the ass-spaghetti-stinky piece of code that sometimes we have to write to go on with our lives.

I hated the idea to have some ugly loop (since I discovered linq, all loops are ugly!) that iterated 30 times and grabbed the current date the chart was accessed and then subtracted the index of the loop, and finally made sure that the result was sorted ascending.

So Range function and LinqToXml came to the rescue! Actually, I didn't know that Range() was available in .Net. A co-worker pointed out to me that I did exist and that it was under the "Enumerable" Class. (Thanks Connor!)

So, the potentially "ugly" loop ended up being a beautiful one-liner linq query:

Dim category = (From index In Enumerable.Range(0, 30) Select _
<category label=<%= DateTime.Now.AddDays(-index).ToString("M/dd") %>/>).Reverse

Man, Linq rocks! I can't stop saying it. Well now that I had the hard part done, I had left to generate some empty data sets and then put it all together:

Dim emptySets = From index In Enumerable.Range(0, 30) Select <set value="0"/>

'Build the Xml template
Dim chartData = <?xml version="1.0" encoding="utf-8"?>
<chart>
 <categories>
  <%= category %>
 </categories>
 <dataset SeriesName="Name1">
  <%= emptySets %>
 </dataset>
</chart>

'return the build template to the calling function
Return chartData.ToString

So the ouput of my query would look like this:

<chart>
  <categories>
   <category label="6/26"/>
   <category label="6/27"/>
   <category label="6/28"/>
   <category label="6/29"/>
   <category label="6/30"/>
   <category label="7/01"/>
   <category label="7/02"/>
   <category label="7/03"/>
   <category label="7/04"/>
   <category label="7/05"/>
   <category label="7/06"/>
   <category label="7/07"/>
   <category label="7/08"/>
   <category label="7/09"/>
   <category label="7/10"/>
   <category label="7/11"/>
   <category label="7/12"/>
   <category label="7/13"/>
   <category label="7/14"/>
   <category label="7/15"/>
   <category label="7/16"/>
   <category label="7/17"/>
   <category label="7/18"/>
   <category label="7/19"/>
   <category label="7/20"/>
   <category label="7/21"/>
   <category label="7/22"/>
   <category label="7/23"/>
   <category label="7/24"/>
   <category label="7/25"/>
  </categories>
  <dataset SeriesName="Name1">
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
   <set value="0"/>
  </dataset>
</chart>

So, there it is. no messy loops, no smelly code, just ....Linq.
Hope this helps

Consuming an RSS feed using LinqToXml

I needed to display some RSS feeds into a dashboard page in one of our websites. I figured I had to mess with XML, XPath, etc.. to browse the xml structure of the post. However, LinqToXml turned out to be the perfect tool for this task. I used Scottgu’s post as a reference, but did my own little twist at the end to avoid another “loop”

First, pass the Rss feed Url into an Xdocument variable:

Dim feedXml As XDocument = _
XDocument.Load(“http://feeds2.feedburner.com/AppriverMalwareWatch”)

Now I need to iterate through the “item” children tags in the XML document and get the “link”, “title” , and “pubDate” elements. Also, I am going to display each post as a hyperlink in an unordered list with the date next to it. Finally, I only need the most recent 8 posts. Here is where magic of LINQ comes in handy:

Dim myFeeds = From feed In feedXml.Descendants("item") Select _
<li>
 <a target="_blank" href=<%= feed.Element("link").Value %><%= feed.Element("title").Value %>
 </a><%= Date.Parse(feed.Element("pubDate").Value).ToString("M/dd") %>
</li> _
Take 8

'Add xElements to the String literal
Dim myUL = <ul>
             <%= myFeeds %>
           </ul>

Now that I have added the feeds to my unordered list, I need to write the html to the div container:

'Write string literal to Div's body
Me.uxDivUls.InnerHtml = myUL.ToString

That's it. In a few lines of code I was able to Consume an Rss Feed. What is there not to love about linq to xml??

Happy coding!

Chicken Chicken

ClientQueryString Page Property

I was browsing through the Page class in System.Web.UI assembly and I run into the "ClientQueryString" property. Here is its signature:

Public Readonly Property ClientQueryString As String.

It returns the page's querystring in an encoded fashion (so one would have to use
HttpServerUtility.UrlDecode to decode the querystring.) Inspecting the source code I did noticed a couple of interesting things: The property goes through the Page.s_systemPostFields and adds them to a "black list" which will be excluded from the string returned by the property. Also, and I think pretty neat, is the way the black list gets excluded from the query string results: The Request.QueryString gets casted as an HttpValueColletion object and then it invokes its overriden "ToString" Method to which the "blacklist" is passed to in its arguments. (excludeKeys is a Hashtable of "keys to exclude".)

Me._clientQueryString = DirectCast(Me.Request.QueryString, HttpValueCollection).ToString(True, excludeKeys

I thought that was pretty cool, no need to loop through the collection, just pass it as an argument and the .ToString takes care of it.

With this property you can't get an individual querystring value since you can't specify a "key name". You get all of query string values. I guess this is useful if you need to pass on the querystring to another page or if for some reason you need perform some custom parsing and manipulation on the query string.