We've had quite a bit of reshuffling in the Java IDE tools this time as compared to last year. The most preferred IDE for the Java platform this time is Eclipse, having a comfortable lead over the rest. Eclipse was released to the open source community early last year. It entered our Developers' Choice Club at the number four position last year, but it's quite surprising to see it rise on the popularity charts in such a short period of time. Sun's Java Studio has dropped to number two, from being the winner last year. Microsoft's J++/J# has swapped positions with Borland's JBuilder this time. Interestingly, the ranks were almost similar for the question where we asked the respondents about their primary Java IDE. The only difference was that Borland swapped places with Microsoft and IBM VisualAge disappeard.
Surprisingly, once again we saw IBM's Visual Age get plenty of votes, despite the fact that IBM discontinued the product long time back and recommended a migration to the WebSphere Studio Application Developer. We had some votes for WSAD as well, but possibly the brand recall for VisualAge is very high due to which it's still coming up. We've got another open source tool enter the club this year, NetBeans. Going by how Eclipse became so popular in such a short period of time, are we going to see the same thing happen with NetBeans? We'll see that next year.
We also tried to figure out whether any particular platform is gaining more popularity as compared to others. For this, we asked whether the respondents were using any other platform two years ago. What we found was status quo for everybody except Eclipse, which has gained its market share. It's also visible in the results.
Like last year, only two platforms have emerged as the key ones across the developers we surveyed, and these are Java and .NET. Once again, Java continues to be the winner for the second year in running and has a comfortable lead over .NET. Like before, we first asked the respondents which all development platforms were they working on. In this question, they were allowed to give more than one option. So basically, a company could be using more than one platform for its development. The graph shows the results of this question. We also asked the respondents for their primary platform for development. Here again, Java emerged as the winner (see the table), and the gap with .NET in this was even greater compared to the first question.
It seems that across the people we surveyed, affinity towards Java as a platform seems to be higher as compared to .NET. Wherever Java was the primary platform, it got higher votes across companies of all sizes, whether they were having up to 50 developers or more than 500 developers. Of the responses where we asked for multiple options, the gap was a little lower. For companies with up to 50 developers, Java and .NET got equal preference. The gap was also quite narrow amongst respondents from companies with 51 to 100 developers. Amongst the respondents working for larger companies, the gap between the two was much wider. There were other platform choices also given by people, but their numbers were too insignificant to include in the survey. Nevertheless some of the other names we got from respondents were VB 6, PHP, and even C++. These were also considered as platforms by them, although their percentages were fairly low.
IBM Rational Rose steals a march on the rest of the competition with a 74 UPI point lead over its nearest rival, Mercury Interactive's WinRunner. Last year, WinRunner was not a candidate at all- this time it enters at the number two slot, pushing last year's runners up down by one step and 9 UPI points. CA Erwin thus, drops to third and Win Case to fourth. Borland Together Control Center remains at last place in the Developers' Choice Club. IBM is winning this for the second time in a row, since we instituted these awards last year. Borland is perhaps the only stagnant player here, since they retain the 10 index points they had in last year's survey.
CASE stands for Computer Aided Software Engineering and refers to all the tools used by a developer in developing and finalizing their application development. Lately, the term 'application development' has come to refer to a diverse area, from client/server, desktop, web, mobile and database applications. RDBMS systems are now including the ability to add scripts and program-lets within themselves and as a result, CASE tools include database tools as well.
A majority of those using Rational Rose as one of the CASE tools in their development effort included of course software houses. But there were also a good number of IT educators and business process consultants in this number. When asked which tool they used primarily in their company, 10% of the respondents picked IBM Rational Rose and Enterprise Architect and put them on top. 24% of the respondents said they did not use a CASE tool. Compared to last year's five candidates, this time we had 29 total entries in this category as an indicator that several companies have now started using some CASE tool or the other.
Primary .NET IDE
We have a unanimous winner here, just like last year. Visual Studio.NET continues to be the winner second year in running with no other competition in sight. This was the only category in the entire survey where we had only one winner. Whether we asked people about which all .NET IDEs they used or the primary .NET IDE, the choice was always one.
The only other responses we got were for Borland Delphi.NET, Visual Studio 6, and Visual Basic 6. However, the numbers for these responses were too insignificant.
This time, we also added another question to the survey. In this we asked the respondents whether they had shifted to their current IDE from some other one, two years ago. If they had, then we wanted to know which IDE were they using earlier. While a majority of the respondents had been using Visual Studio.NET in the past as well, we got a few responses for VB and Visual Studio v6, which are again from Microsoft.
Only about 6% of the respondents said that they were using Sun's Java Studio.
BugZilla, available under the Mozilla Public License remains the favorite bug-tracking tool for the second year in a row. But it had better watch out, since others are closing up, fast. We must point out here that the existence of separate bug tracking and debugging tools has made this a densely fought category with 32 candidates in total. Mercury's WinRunner and LoadRunner shoot to second place just 2 UPI points behind BugZilla and IBM's Rational Robot and PurifyPlus slot into third four points short of the winner. Compuware QA Center completes the Developers' Choice Club in a rather distant fourth. Last year, quite a number of developers were using custom tools, putting that set into second place with 28 UPIs.
This time however, we see a different picture, with only 2% of the respondents selecting that option. Also, in 2004 we saw Visual Studio in the second spot, standing in for IDEs. This time, IDEs seem to have fallen out of favor with standalone debuggers and bug trackers being the only entrants. We saw about 20% of our respondents select the 'None' option, indicating perhaps that they had no defined system in-place to handle these tasks.
The Developers' Club for this category remains exactly the same as last year, meaning that for the second time in a row, WinRunner and LoadRunner sweep victory, although by a smaller margin than last time. They are now just 17 UPI ahead of second placed Rational Robot and PurifyPlus (the gap was 54 points last year). Compuware QA Center manages to remain third, but loses four UPI points.
In 2004, manual methods of testing was the third most preferred, in a tie with QA Center. This year, those methods are losing favor-manual methods are now in the last place, with 12 UPI points. Telelogic's testing tools gain fourth place and five more points over its 2004 showing.
Yet another category where about half the respondents (46%) said they did not use a testing tool. A good percentage users of the software that made it to the Club agree that they use it as their primary tool as well. Again, standalone testing tools are in favor over integrated environments across the board. A lot of these tools are also cross-platform, indicating that cross-platform development is also on in a big way.
The top three from last year remain at their spots, with changed scores. Microsoft Visual SourceSafe wins in the category. But IBM Rational ClearCase gains 53 UPI points from 2004 and CVS gains 57 UPI points. There is a new entrant into the Developers' Choice Club this time in the form of Linux Sub Version.
Linux Sub Version is an up and coming replacement versioning tool for CVS and seems to be rapidly finding fans across the board. For the second year in a row, VSS gained 40% of the votes.
Rational ClearCase and CVS both had comparable shares with 38% and 33% of the votes respectively. The only other significant vote-share was the number of companies with no versioning setup.
CVS has a higher number of companies using it as the primary versioning tool compared to ClearCase which sees it at the number two spot in that count. If we combine the votes for CVS and LSV, it actually becomes the winner in this category, beating VSS by three UPI points.
We suspect that given the rise in popularity of CVS and its replacement by LSV in the past year, we should expect to see this combo in a higher position next year.
There is a change in the guard at the top step. Oracle wins this year, pushing Microsoft SQL Server down to second spot. MySQL and DB2/Informix stay at their last year's third and fourth spots respectively. And there is a new entrant to the Developers' Choice Club 2005, in the form of PostgreSQL. Both MySQL and DB2/Informix leap-frog ahead in terms of UPI-MySQL gains 30 points and IBM takes 13 points. PostgreSQL has four points more than the final Club member last year (IBM with 18).
The primary RDBMS in these companies remains more or less the same, with PostgreSQL dropping out of the 'club' in that case. RDBMS is a category where software development companies have a lot of influence in databases purchased by users. Of the people using these databases, Oracle and SQL Server votes mainly came from ERP solution users and software developers. Also interesting is that a big chunk of both users are in the sub Rs 50 Crore turn over segment. Government and Manufacturing are the two key verticals where they are strong. This is a healthy mix of platform specific and cross-platform databases.
This category sees IBM WebSphere capture top honors. Tomcat comes second with J2EE entering the Club in the third spot. Last year's winner-Windows Server 2003-is fourth. BEA WebLogic, JBoss, Oracle 10g AS, Macromedia JRun/ColdFusion and Borland Enterprise Server complete the Club for this year. We see a lot of places changing hands from last year. WebSphere was in the second spot last year with Tomcat in third, WebLogic in fourth and Oracle in fifth.
The gaps have also closed up between the players. Last year, we saw Win 2003 winning by a 65 UPI point margin and this year's winner is just 10 points ahead. Even among the rest of the pack, the gaps are within ten or twenty points each-an indication of more vigorous competition among them on the ground. Of course, the majority of those polled were software developers.
Also interesting is the picture that emerges from combining our Users' Choice and this survey's results. In this category, users have voted Oracle 10g as their preferred app server with a 27% inward brand-shift for next year. WebSphere that wins here came second there, with a 23% inward shift indication in the Users' Choice.
These are tools that let developers convert their code into installable applications. InstallShield remains the most popular of the packaging and deployment tools, with Microsoft Packager (Windows Installer) being the number two. Last year's number three, NullSoft's Installer remains at third but sees a tie with SoftwareShield System. Compared to last year, the gap between InstallShield and the second-placed tool is larger by 20 points more. Similarly, NullSoft sees a UPI drop of three index points from its tally of 11 last year. Commonly visible tools like ones from Wise, StuffIt and zipper application vendors don't make it to the club-Wise and StuffIt were candidates in this survey though.While a large majority of those polled indicated they used
InstallShield as one of their packaging tools, this number dropped significantly and the gap between InstallShield and the Microsoft tool drops to just 10 index points. In that club though there is a new face-Install Anywhere which isn't even in the first club and scores a lonely 16 index points to just about make it in there. And NullSoft isn't among the top three in those votes.23% of respondents said they did not prefer any one tool as their primary tool for packaging and preferred with what suited a particular need. This is also one of the rare categories where we did not see anyone pick the 'Custom' option.