John Lawrence (MSFT)

Final version of TFS User Administration Tool now available

johnlawr — Thu, 18 May 2006 08:27:00 GMT

Joe Morel just contacted me to let me know that the final version of the Team Foundation User administration tool that I've blogged about in the past is now available for download, running against RTM TFS.

Here's a quick description of the tool

The TFS Admin Tool allows a TFS administrator to quickly add users to all three platforms utilized by Team Foundation Server: Team Foundation Server, Sharepoint, and SQL Reporting Services, all through one common interface. The tool also allows administrators to change the current permissions on any of the three tiers, identify any errors, and view all of the users and their permission sets across Team Foundation Server, Sharepoint, and SQL Reporting Services.

Here's a link to the project homepage

Template project for subscribing to team foundation events & notifications

johnlawr — Sun, 07 May 2006 16:37:00 GMT

A couple of people have contacted me in the past wanting more details on how to subscribe to the various events that we expose in TFS. Currently our documentation/SDK probably isn't as detailed as it will ultimately reach in this respect.

However, Howard Van Rooijen (of Conchango's Scrum For Team System fame) has managed to put together a template project which creates SOAP endpoints for our core notifications and even performs basic deserialization.

Here's the link. Team Foundation Server Notification Web Services: Visual Studio 2005 Project Template

This is very cool - I'm sure it will be extremely useful.

MindManager integration with Team Foundation Server for requirements gathering

johnlawr — Tue, 18 Apr 2006 17:03:00 GMT

A couple of weeks ago we had a Dev Lab on campus - during which time Michael Scherotter from MindJet was campus working with a few people from the TFS team who were showing him some of our Integration Features.

Mareen Philip forwarded a video to me of the end result - this is what Michael came up with after just four days - full integration with MindManager for requirements gathering synched up with TFS.

You can download the Mindjet Requirements Manager for TFS here and you can download a shockwave video demo here (video is also available inline on the page here)

Michael wrote a blog on his experiences here. It's well worth a look to see both the end results (a very cool extension to Mind Manager) as well as how much potential there is in integration with TFS.

What other ideas do you have for TFS integration solutions?

Jason - another Team Foundation Developer - is blogging

johnlawr — Thu, 23 Mar 2006 08:22:00 GMT

Looks like Jason Prickett is now blogging.

I'm looking forward to reading what he has to say. He's one of the developers on my team who has been responsible for much of the Work Item Tracking user interface elements of the Team Foundation client pieces.

He's also been working on the Team Foundation power toy for command line interaction with Work Items that I asked you about a week or two ago. I'm not sure when we'll see it launch yet, but he's used a lot of your input during his implementation.

Blog on, Jason!

Too much email, not enough life

johnlawr — Wed, 22 Mar 2006 05:46:00 GMT

Via Itzy I came across this poem from Steve Palvina.

Will you invest your time in what really matters to you, or will your tombstone ultimately read like this:

Here lies John, who passed away
While answering his email one day.
No friend, no child, no loving mate
Could keep poor John from working late.
With each new mail, he worked like hell
To click ”reply” instead of “del.”
A prompt response he’d always give
But somehow he forgot to live.

Ouch. Sounds horribly like me.

We did have green beer when Team Foundation shipped

johnlawr — Tue, 21 Mar 2006 08:44:00 GMT

A few people noted that Mary Jo Foley thought we'd be having green beer when we signed off on Team Foundation Server on St Patrick's day.

Here's the proof - Etienne and Aaron enjoying a quite revolting looking drink... :-) It tasted good though.

Team Foundation ship room signs off on RTM - ship it!

johnlawr — Fri, 17 Mar 2006 16:34:00 GMT

We just finished our final shiproom - Team Foundation Server is ready to go! Read more on Jeff's blog here

These are just a few of the folks who have been at shiproom every day for the last few months helping get this baby out of the door - thanks to all of you!

Who's who?

From left to right - back row: Dan Kershaw, PM; Mohammad Iqubal, Test Lead; Aldo Donetti; Gregg Boer, PM; Jim Boyle - PM, responsible for running Ship Room; Paulo De Oliveira, PM; Marc Kuperstein, PM Setup; Jeff McKune, Dev Lead

Front Row: Jeff Beehler - Jiminy Cricket of TFS; Bryan MacFarlane - Dev Lead; John Lawrence (me) - Dev Manager; Layne Wiwatowski - Test Lead

You can bet there will be more celebrations throughout the day...

What features would you like for a command line work item tracking tool?

johnlawr — Mon, 06 Mar 2006 21:55:00 GMT

One of my team is looking into what a command line tool for Work Item Tracking might look like. We'd like it to be something that you could use without a GUI so that it could be suitable for scripting.

What would you like it to do? What problems would you use it to solve? Does it sound useful to you?

I'd love to have your input - I want to make sure we're creating something that is useful to you. Feel free to post a comment here, or contact me directly.

Thanks!

A fellow email blood pressure sufferer

johnlawr — Sat, 04 Mar 2006 09:10:00 GMT

I just read David Anderson's post on how he just decided to Take Back His Life after his email inbox caused his email blood pressure to skyrocket.

David, you were getting 400 emails a day at the peak of the Beta 3 period for Team Foundation Server. I apologise - I was probably the sender of a large number of them...

I know exactly how you feel - in fact if I only received 20 emails an hour I think I'd be happy. I started getting this quantity of email when I first joined the team and rapidly realized I wasn't going to be able to cope using my previous strategies. Thanks to some advice from Chris Shaffer, our Test Manager, I discovered Getting Things Done and managed to get - and keep - things under control.

As a process for managing mail, this worked well - but for in parallel for a few years I have felt like I've been fighting with Outlook. It would take 10-15 minutes to open up and become responsive, and every so often would start giving me errors telling me it couldn't open or display folders because it was out of resources.

Two weeks ago I finally cracked. It reached the point where this was happening all the time. It became more and more difficult to manage my email and get things done - compounded by the fact that these issues had stopped autoarchive and email rules working, so my exchange mailbox reached it's max size and I couldn't even send mail any more. This was the point that my email blood pressure finally cracked - I felt more stress than I've felt in years.

I could no longer manage my email. And no-one knew why. We are lucky here to have a great internal support system - we first assumed that it was my client, but after reproing the issue on 4 machines (and baffling a friend of mine on the Outlook team) finally realized that we'd have to rebuild my exchange store.

After a week of basically being incommunicado, Outlook was back, I caught up - and I am so much happier. Outlook actually really performs quite well indeed and I now realize that for the past several years I've been working with one of my primary tools working at about 20% efficiency. If only I'd discovered this sooner, I wonder how much better my work/life balance might have been.

Some history and clarification on our Team Foundation dogfood statistics

johnlawr — Fri, 03 Mar 2006 20:32:00 GMT

I've been really bad at reviewing comments on my blog lately - sorry. (I'm sure you'd rather I was focusing on shipping TFS though...)

I just came across this comment from Tim Rowe who was wondering how we reached 600 dogfood users for our November stats.

When it says 'Team foundation', is that 600 people working on JUST the Team Foundation server portion of VS?

That's a massive teams for such a... uh... well, I wouldn't exactly call it a massive project. Nothing I'd expect to see 600 people on anyway.

It would be interesting to know how many of them are actually programmers, and how many have 'other' jobs - eg, testers, architects, managers etc who are commiting 'non-code' files.

Clearly some absolutely crazy restructuring happened here between the weeks of 24/6 and 24/10 (from ~70,000 files up to 600,000 files... on that alone maybe I can retract my first statement).

600 users is not just Team Foundation. We do have a fairly large team working on TFS - but certainly not that big. We share a dogfood server with a number of other teams product groups. Over the past couple of years we've been steadily increasing the number of people, teams and Team Projects hosted on the server.

Way back in the very early days of development, the Work Item Tracking team ("Currituck") were dogfooding work item tracking on their own server, and the Source Control folks in North Carolina ("Hatteras") were dogfooding version control.

We then moved onto a shared dogfood server, and when it was working tolerably well we expanded the use to include other members of the VSTS suite (folks from the Team Architect, Developer and Tester organizations).

Over time we added more teams from across Microsoft. Gutsy folks, who wanted to be at the bleeding edge of software project management tools... Some of these were small projects - others much bigger. All of these added more and more folks to the number of users in the system.

In fact the number of users of the server is higher - the count of 600 is the number of folks with assigned work items - there are more users who connect and run queries, look at bugs - without having them assigned to them.

To address another comment from Tim:

Clearly some absolutely crazy restructuring happened here between the weeks of 24/6 and 24/10 (from ~70,000 files up to 600,000 files... on that alone maybe I can retract my first statement).

Every team also brought more code with them - some smaller code bases, others significantly bigger. However, one of the sources that has boosted the total number of files in the system has been when we've brought in more of the entire developer division source tree into our dogfood server. Sometimes this has happened when we pull in a new branch of large portions of the code base.

Another earlier example of file growth was when Brian first decided he wanted to pull in all of the .Net Framework code into our dogfood server for the first time (to start with we just had the VSTS code on the server). This made a big difference to our file count. Brian called me me up to tell me he was doing it one afternoon during a dry run of a presentation Doug Neumann was going to be giving at PDC. I remember watching the server through the session, initially nervously, but later very happily as I realized we really were going to be able to handle all the checkins...

It's been a great journey - and I'm very grateful to everyone who's been prepared to eat our dogfood for us. It's been painful for them at times, but ultimately we've ended up with a much better product in the end. Dogfooders - thankyou!

Team Foundation User Admin Tool now works against the TFS RC

johnlawr — Fri, 03 Mar 2006 18:16:00 GMT

Some time ago I blogged about a tool to manage users and groups across WSS, Reporting Services and Team Foundation.

This has been available in Beta in the past, but only running against old builds. The Developer Division Aftermarket Solutions team has just updated the binaries to run against the Release Candidate of Team Foundation.

They blogged about the update here - other links for the tool are:

GotDotNet Site (including download links): http://go.microsoft.com/fwlink/?LinkId=59385VS
Power Toys Forum: http://go.microsoft.com/fwlink/?LinkId=59386
Bug Tracking Site: http://go.microsoft.com/fwlink/?LinkId=61472

Note - this should work against the final bits we release at RTM - I'm 99.99999% certain we won't take any last minute bug fixes that would break compatibility between RC and RTM.

What other Team Foundation tools would you like?

Team Foundation Dogfood Statistics - February

johnlawr — Thu, 16 Feb 2006 16:45:00 GMT

It's not long since I posted the last numbers, but Brian has just sent around the latest information, so here it is.

Some key points that Brian highlighted:

Number of active users is 587 - our target for shipping was 600, so we're close enough to declare victory here!
7 day availability for the TFS server is 100% – We’ve been real close before but it’s fantastic to see that we’ve had no down time in the past 7 days! The 30 day availability is 99.92%. These numbers, however, exclude warehouse availability - with the reporting warehouse being offline for a long period while the SQL team helped us debug some issues.

The big spike above was when we migrated a large new team onto our server - this week we're back to the regular pattern.

Users

Recent users: 587 (up 70)
Users with assigned work items: 795 (up 115)
Version control users: 864 (up 155)

Work items

Work items: 67,004 (up 5,000)
Areas & Iterations: 2,755 (up 300)
Work item versions: 574,365 (up 42,000)
Attached files: 19,952 (up 1,500)
Queries: 6,979 (up 850)

Version control

Files/Folders: 1,507,247/178,953 (up 310,000/40,000)
LocalVersion: 48.9 M (up 12.9 M)
Total compressed file sizes: 95.2G (up 10.5G)
Workspaces: 1,770 (up 300)
Shelvesets: 2,458 (down 400)
Checkins: 32,573 (up 13,000)
Pending changes: 57,278 (down 15,000)

Commands (last 7 days)

Work Item queries: 18,296 (unchanged)
Work Item updates: 7,875 (down 18,000)
Work Item opens: 28,936 (down 20,000)
Gets: 7,908 (up 200)
Downloads: 5.3M (up 2.6M)
Checkins: 1,032 (down 2,800)
Uploads: 7,409 (down 75,000)
Shelves: 649 (up 200)

Team Foundation Dogfood Stats - January

johnlawr — Tue, 31 Jan 2006 18:40:00 GMT

(Update - this didn't get sent to the main aggregate site when I first posted it)

Brian sent around the latest updates in our dogfood usage as of 1/10 - I've been swamped so it has taken me a while to get around to posting them. (I'm also trying to write a Team Foundation report to generate some of the charts - not all of the data is available in the warehouse by default so some of the charts aren't so easy to create).

This report represents a significant milestone. In the week covered we've had 517 distinct people use the dogfood server. This is the first time weve broken 500 and a new record by a pretty wide margin. Youll also notice a huge spike in checkins this week. This is due to a new team that have started dogfooding on the server who have been doing 3,131 checkins as part of a conversion process.

Another notable statistic is that the server availability is getting consistently better. We hit a low on 30-day availability in late November of around 92% (mostly due to Warehouse issues). Today our availability stands at:

7-day: 99.66%
30-day: 99.84%
Our recent numbers have taken a bit of a hit due to a problem with password resets and a couple of OTG OS patch induced outages. Overall the availability of the core TFS services is very high.

Users

Recent users: 517 (up 82)
Users with assigned work items: 681 (up 120)
Version control users: 707 (up 102)

Work items

Work items: 62,066 (up 7,000)
CSS nodes: 2,434 (up 250)
Work item versions: 531,941 (up 55,000)
Attached files: 18,542 (up 4,000)
Queries: 6,113 (up 700)

Version control

Files/Folders: 1,193,463/139,198 (up 250,000/30,000)
LocalVersion: 36.0 M (up 6.5 M)
Total compressed file sizes: 84.5G (up 10.5G)
Workspaces: 1,470 (up 200)
Shelvesets: 2,458 (up 200)
Checkins: 19,627 (up 6,000)
Pending changes: 72,173 (up 33,000)

Commands (last 7 days)

Work Item queries: 18,378 (down 3,000)
Work Item updates: 25,935 7,401 (up 18,000)
Work Item opens: 49,353 (up 2,000)
Gets: 7,728 (up 1,900)
Downloads: 2.7M (unchanged)
Checkins: 3,851 (up 3,400)
Uploads: 83,837 (up 30,500)
Shelves: 458 (up 60)

Upgrading to Team Foundation RC: HTTP Header SOAPAction errors

johnlawr — Tue, 31 Jan 2006 17:55:00 GMT

Overnight last night we upgraded our dogfood server to the latest build of Team Foundation prior to our upcoming Release Candidate.

The upgrade went pretty smoothly and we're all back up and running this morning. However, users of old clients are hitting an issue which I though might be worth bringing to a wider audience.

The error you'll see is "Server did not recognize the value of HTTP Header SOAPAction"

In this Release Candidate we have upped the version number of our web services, because beta clients are not compatible with the final RC/RTM server. You might see this if you upgrade your server to the RC build but some of your users don't update their clients. The solution is to uninstall Team Explorer, and reinstall the RC or RTM Team Explorer.

The worst case of this will be if you've got a solution with multiple projects bound to source control - you'll see one of these error boxes for each project in the solution.

This is the last time we'll do this. Barring some major unforseen issue cropping up, the RC clients will be compatible with the final RTM server (although obviously you'll want to upgrade clients to the final bits when they are available).

More about the impact of clicking Send Error Report

johnlawr — Fri, 09 Dec 2005 22:22:00 GMT

I blogged last week about how important it is to us for you to click "Send Error Report" when the Windows Error Reporting dialog boxes appear. It's probably changed now, but much to my amusement I discovered that yesterday if I simply typed "Send Error" into MSN Search or Google that my blog entry was the first hit!

I've had a few comments that I wanted to respond to, and I have some more data to share that Jeff and Jiange just sent me today.

More detailed data on how have we resolved the error reports for Team Foundation so far
Jeff and Jiange did some in mining to correlate the number of "Watson hits" with the resolution of the corresponding bugs. Every time someone clicks "Send Error Report" counts as 1 Hit. If you have the same crash and send in a report in multiple times, or multiple users hit the same crash, these will all (usually) correspond to a single Watson entry in our database (and a single bug). Sometimes we discover that we have opened more than one bugs to track hits which end up being the same underlying issue, so these show up as resolved as "duplicate" instead of "fixed"

The important observation from the data we have so far - is that according to the table below we're actually fixing 99% of all Watson hits for Visual Studio Team Foundation. Obviously this isn't a guarantee that you won't run into an issue that we haven't found or fixed yet, but I'm really happy to see that we've spent our time fixing the right issues so far.

Resolution	Count	Percent By Count	Hits	Percent By Hits
Active	5	2%	5	0%
Fixed	152	65%	250886	86%
Duplicate	36	15%	38627	13%
Won't Fix	18	8%	904	0%
Not Repro	18	8%	819	0%
By Design	1	0%	3	0%
Postponed	3	1%	15	0%
External	1	0%	21	0%
Total	234	100%	291280	100%

Please read my earlier post to see what I found out when I checked to see if we made the right calls when we didn't resolve a bug as 'fixed'.

Can I use Windows Error Reporting in my own apps?
Alfred Wallace asked if there is a way he could use the Windows Error Reporting technology in his own applications. I haven't really looked into this too deeply myself, but we do offer this service to 3rd party developers from the Windows Quality Online Services web site. This page describes how you can start using Windows Error Reporting logs for yourself. I think I've heard talk of making this even more accessible in future, but I don't have any idea what that might look like.

Why don't my bugs get fixed?
Peter Richie commented that he didn't feel that the error reports he sent over the years for Visual Studio 6.0 had been addressed at all. I can appreciate how frustrating this can be - I've experienced this exact same scenario with other Microsoft products in the past (and some fairly recent ones too, but I'll refrain from naming them here). It's definitely true that different teams seem to be able to get higher fix rates than others for these "Watson" issues.

I don't know if it will help you Peter, but I tried to think through some reasons for why you might have observed and I think there are several. The first thing that springs to mind is that Visual Studio 6.0 is an old product now - and, as you know, is no longer under active development. Once we've shipped any product, the cost for delivering fixes for any bugs increases astronomically and in the past we haven't really had a great vehicle for shipping fixes for Visual Studios issues that we might not call "recall class". I'm really proud of our fix rate for issues we found in the Beta program, but I know that this will drop for issues we don't uncover until after RTM.

Visual Studio 2005 is the first version that we'll ship a service pack for - in the past we haven't done this, so the bugs have gone unaddressed. It's quite possible that your bugs were addressed in subsequent versions of Visual Studio - if you're still seeing the same crashes in Visual Studio 2005 that you saw in Visual Studio 6.0 please let me know and I will chase down the issue for you.

Secondly, the bugs were most likely to be addressed during the earlier years of Visual Studio 6.0's life span - but at that time the tools for handling Watson reports were really limited. Our infrastructure for handling Watson reports has been significantly improved over the years (take a look this entry on Chris Pratley's blog for some history). Today there are some really sophisticated tools to help me gather data, debug in the dump reports that get sent in to us, and even refine the data we collect for specific crash reports over time. Unfortunately, this probably wasn't available to help debug your VS6 issues.

I know that this won't satisify your frustration over these older bugs but I hope it illustrates how we're taking steps to do much better in future.