Couple of PDC 2009 thoughts too big to fit in 140 characters – praise to Microsoft

PDC 2009 is over and what a conference that was… For folks there tablet for free, ton of excitement experienced first hand and a lot of valuable social networking… For us others not being able to attend there’s a ton of prime time learning material I am really looking forward to see.

In general, amount of positive vibes from community was so high that even surprised the usual MS haters so we had to wait couple of days to start with their choir attacks on Microsoft. I don’t work for Microsoft, I am not an MVP in fact I don’t have any relationship with Microsoft but I still find as not fair a bunch of things I hear considering how much investment we see from MS in the development space and thus I wanted to speak in Microsoft favor in a very opinioned manner … Not claiming things in this post being universal truths – they just reflect my takes on various subjects related to Microsoft development ecosystem.

The glass is half empty…

“Microsoft Silverlight 4 sucks because it is not any more cross platform orientated”

Where it all started: http://www.theregister.co.uk/2009/11/20/silverlight_4_windows_bias/.

One opinion I agree with in general: http://blogs.silverarcade.com/silverlight-games-101/21/silverlight-is-silverlight-4-moving-away-from-cross-platform/

Here are couple of thoughts of my own on that subject:

  • I never heard anyone from MS saying that SL 4 goodies won’t be supported on Mac.
    Judged on a size of Windows (4 Mb) and MacOsX (20+ Mb) Silverlight installers, Silverlight on those two platforms is probably different code base with the same API. Even if not, I can imagine thadst a custom Mac implementation of clipboard access, web cam etc can be done using custom coding for Mac OS API.
  • Does Microsoft really need to spend its own money for Mac users?
    In my personal opinion not really because of couple of things:
    • Number of users with MacOS is insignificant from market share perspective so ROI is not as high as for the investments made for Windows users.
    • Even with that low market share investing in MacOS might have sense only if Mac zealots are not so brainwashed and hate everything from Microsoft. Just check out their comments on how they despise the sites asking them to install “Microsoft bloat ware called Silverlight”, how they would sooner cancel their Netflix subscription then install Silverlight to watch HD
    • Before Windows 7, Microsoft maybe had to play on “works on Mac too” card to reduce the impact surface for the Apple ads, but now with Windows 7 premier OS they don’t have to. Table got turned and IMHO it is now in Apple interest to support Silverlight on Macs because in couple of years it would be everywhere and their users would simply have to have it in other not to be cut of from the most of internet sites. I really believe in that.
  • What about Linux?
    Well, Moonlight development is in Novel hands anyway so it would be up on them to determine the speed when this would come. Considering the fact that MonoTouch looks like their primary interest I don’t think it would happen soon which considering the reasons similar to Mac (low market share, blind hate toward anything coming from Microsoft) is (IMHO) not a big deal.

"Microsoft Silverlight 4 would break the security model of Windows 7”

So, the pitch of this attack is that due to the fact that Silverlight now provides “full access” mode of sandbox,  we are back basically to the ActiveX era where in order to run a web site (which you want) you have to allow elevated rights to that app. In other words, they said that everyone would “OK" that and the hell gates would be open by that.

Here’s a good blog post summarizing what SL 4 full trust mode means for real (http://mtaulty.com/CommunityServer/blogs/mike_taultys_blog/archive/2009/11/18/silverlight-4-rough-notes-trusted-applications.aspx)

And here are couple of thought of mine related to this subject:

  • The biggest difference between COM and Silverlight is that Silverlight application running in browser is ALLWAYS sandboxed. In order for user to elevate sandbox rights Silverlight application has to be INSTALLED on desktop (which is first safety switch not a lot of users would pass) and during installation there is clean warning informing users about implications.
  • It is “elevated” not “full”. One example (as Mike explained in his blog) is that even in elevated trust you still can not access any folderfile on user hard drive.
  • I have a feeling that the same guys praising Adobe AIR for its tweeter clients never questioning attack surface AIR has due to ITS own full trust and mocking Silverlight for its too secure sand box preventing developers to do cool apps are the same bragging about the sandbox.
  • Microsoft had to listen loud community demands for cross domain web service calls, enable premier full screen experience for media players and kiosk applications.
  • Enabling access to COM in this trusted mode enabled integration with a bunch of software and opened some really interesting and productive implementation ideas. If you would have so much applications of your own including the prime time Office applications, would you seriously pass the card of enabling their integration. I wouldn’t skip that chance for sure 🙂

“Microsoft DeveloperDesigner workflow is just a myth and thus XAML instead of code based approach is wrong way to go”

Well, in a way I agree that right now most of the creative folks are working with Photoshop, Illustrator, Flash, CSS, HTML and that not a lot of them won’t laugh when BlendSL would be mentioned to them as a career option. But that is understandable only short term.

There are couple of facts why I think that won’t be the case anymore in a year or two when the demand for MS designers would grow due to next things:

  • Silverlight reach the 45% adoption rate in such a short time without real commitment and investments from industry. That number would just grow with continues investment  and alliances Microsoft is making, so rather sooner then later the tipping point would be reached.
  • Silverlight code is the same .NET which enables reusing of the skills .NET developers have. In other words, I bet good part of millions of .NET developers would step into RIA world in upcoming years which would improve the desirability of Silverlight for big companies.
  • Blend 1 sucked, Blend 2 was ok, Blend 3 was fine (with PhotoshopIllustrator import)… Watching the speed and the trend how RIA tools are improved in MS space one could expect that in a year or two Blend would become very capable tool. If not then, then in next version. The key point is that behind Blend there is MS with endless supply of cache pumping into it on demand.

“Microsoft Silverlight is just next Web Forms”

I agree with this one because I too can see the attempt of MS to bring RIA web development to masses the same they did with
desktop developers in 2002 when ASP NET bring them to web world. I agree and I don’t see anything wrong with it. The whole stateless response/request pipeline was never supposed to handle RIA scenarios we have now so abstracting that with some API (what web forms did) is not possible. Regardless of how many layers you put on top of web, it is web down bellow and sooner or later it would pop up.

That’s why I think MS did the right thing and break up with the web completely. Now the web browser is just used to host sandbox plug-in, and web just to bring to plug-in data required for it work. After that, it is full desktop model application not having anything to do with the web. Yes, you heard me well – Silverlight is for me desktop application deployed through browser and that is good…

“There’s no thing done by Microsoft Silverlight which I can not do with MVC/HTMLx/CSS/jQuery respecting the web standards at the same time”

Well, there are two type of web properties: web sites and web applications. I guess there is no need to spend a lot of time explaining how cost ineffective would be doing web application using jscript RIA approach. Silverlight web application utilizing some of the concepts prism and other framework have allows effective work of multiple development teams on the same site with clean separation of concerns different teams and team members have in that process utilizing server side .NET skills and XAML designer skills. With jscript implementation is lengthy, cross-browser compatibilities crawl their ugly head sooner or later, performance problems with DOM size and different browsers etc, standard .NET developers are not very usable in client side scripting coding etc.. It can be done, but I strongly believe performance, implementation period, maintenance and cost effectiveness would all be on Silverlight side.

As far of standard web sites, until Silverlight don’t solve the SEO problem, I agree that jQuery is better SEO solution for pages visible by prospect user. If your site has a authorized portion requiring authorization then SEO is not so important because crawlers won’t be anyhow able to access it.

“There’s too much magic in RIA service – it is just another demo ware framework”

I looked at RIA services in its early preview version at the beginning of 2009 and sure there is something in the code gen based approach a lot of us would find “not pure” (things might change from there). So what? I can see a lot of normal sized web sites benefit a lot from it. Even the my bellowed NHIbernate works with RIA Services . The way I see it is that they gave as as an option a bunch of prebuilt code we would need to build anyhow in almost every web siteapplication we do. For a lot of folks replacing the total control on your code base in favor of productivity boost is perfectly valid option. At the end of the day, if you don’t like it – don’t use it do it yourself or buy some other framework like IdeaBlade etc 🙂

What could be wrong with free lunch?

The glass is half full…

“Microsoft listen us”

In case you haven’t seen it already go checkout http://silverlight.uservoice.com/pages/4325-feature-suggestions where users voted for features which they wanted to be included in Silverlight 4. MOST OF THEM ARE INCLUDED. We asked, Microsoft responded. Thank you Microsoft!

“Silverlight won’t die.”

Most of the people I follow or read their blogs kind a gravitate toward the feeling that Silverlight is destined to fail, that it doesn’t have a chance against Google backed up HTML 5, that JScript is more then sufficient, that comparing to Flash SL is just a toy…

I admit that recently I started being in doubt regarding my decision to bet my career on SilverlightWPF based presentation layers and even started questioning that decision (bought couple of jQuery books :))

After PDC 2009 I don’t have any doubts and that is totally not related to fabulous presentation Scott Gu did – we all expected from that anyhow so that doesn’t count. For me the much more importance is in the fact that almost every slide in Ray Ozzie Key note on day 1 had something related to Silverlight. As soon I heard him about 3 screens vision and I was sure that Microsoft is betting its future on Silverlight and that they won’t step back in the future and ditch Silverlight. And that makes all the sense on this world and makes me smile.

Another important thing coming from Ray’s vision of 3 screens driven by Silverlight is that I won’t be spending time with the MonoTouch andor other iPhone related technologies. What I am going to do instead is just to wait for Windows Mobile 7 which I’m sure (based on recent great design coming from MS kitchen – ZuneHD last example) would be cool looking and HW rocking (based on Win 7) so I expect to see it well and alive even beside Android and iPhone platforms. If that happens, I’ll just reuse my Wpf/Silverlight skills and start developing for WiMo 7. Another thing I hope to finally be able to do on WiMo 7 ship date is to replace my iPhone with WiMo phone and say goodbye to the Apple proprietary things.

(I’ll still read the jQuery books but I am not going to followread jQuery related RSS feeds)

“WPF is not dead.”

A lot of folks (me included) questioned the commitment Microsoft have in Wpf after seeing the state of SL 4.

I don’t think wpf is going anywhere due to next couple of reasons:

  • Much better development experience in WPF then Silverlight (debugging, tools etc)
  • Visual Studio 2010 is WPF application which shows strong commitment Microsoft has toward further enhancing the WPF.
  • This would sounds weird: WPF has bigger user baseadoption then Silverlight 
    Silverlight is at 45%, WPF is at 90% (every windows XP SP2, Vista, Win 7, Server 2008 has WPF).
    In case you care about windows platform only (like me) WPF is perfect platform to build upon.
  • WPF integrates with OS features (jump lists, progress bars etc), can use sync framework, has direct DB access (for corporate intranets) etc
  • Size of .NET Framework 4.0 Client profile is just 30 Mb (in case of 32 bit version) which is really not a big deal to be downloaded in 2009. In other words, on a PC not having .NET at all, downloading of 30 Mb could make the computer fully capable to execute your app.
    More details here: http://blogs.msdn.com/jgoldb/archive/2009/10/19/what-s-new-in-net-framework-4-client-profile-beta-2.aspx

“Entity Framework is very viable option”

We all remember the EF vote of no confidence which pinpointed the reasons why Entity Framework 1 was not a good option for developers: lack of POCO, DB orientated modeling only etc, etc… Just a year after we are having EF 4 which (judged by PDC 09 Entity Framework session) it looks like they fixed them all. They even work on Code Only (“FluentNHibernate for EF”) feature.

I’m pretty sure that gurus like Ayende would find in lest then 10 seconds 27 NHibernate important features missing in entity framework but to me picking a technology to
invest my time (among other things) is based in highest perceived ROI on my incomes. I had similar dilemma in 1995 when I had to pick between Microsoft Visual Basic 3.0  and Borland Delphi 1.0 and I pick Microsoft VB 3.0. Why? Not because I thought it was better, but because of three things:

  • I estimated that demand for VB in upcoming period would be much higher then for Delphi
  • I estimated that Borland is more likely to loose that fight with Microsoft (who has endless cash reserves)
  • I estimated that amount of knowledge (books, articles, dev community etc) would be much bigger and easier to get in VB case

Almost the same questions I can ask today: NHibernate VS Microsoft EF4  and with a lot of sadness (I really love NHibernate) I have to conclude that EF is the way to go for me because:

  • In version EF4 is at least “good enough” alternative: POCO and model first are possible.
  • Looks like that is Microsoft long term data strategy which means its knowledge would be very valuable in MS oriented development shops
  • Tooling is ok (I don’t buy that designer crawls when you have hundreds of entities because I never have them all thanks to bounded context and it is nice to do all of your development not leaving the VS 2010)
  • There’s a lot of technologies building on top of EF (ADO NET Data Services, RIA Services, Azure)
  • There are some really good booksblogsvideo material already for EF while for NHibernate there are no so much sources of knowledge. Microsoft would for sure outperform NHibernate in this are many times and directly impact the level of adopting,
  • The most important: It would get better and better with every next release. (If they get so far in just a year where they would be next year and the year after the next year?)

I’ve been working last couple of months on my own personal pet project based on FluentNHibernate <–> NHibernate combo, but I am probably going to migrate to EF4 now. If nothing else, I would at least get better understanding on how it stands today for real against NHibernate based development.

Conclusion

I am really happy these days being a developer in Microsoft world seeing so much initiatives, innovations and energy on Microsoft side. I have a feeling that Microsoft was sleeping somehow until a year ago when it woke up and started again doing great things (Bing, Natal, Courier, VS 2010, Windows 7, Silverlight 4, Zune HD etc). No more easy points for Apple adds and no more easy points from haters on things Microsoft was honestly really doing bad mainly due to ignoring the community feedback.

Good work Microsoft!

Advertisements

Posted on November 22, 2009, in Uncategorized. Bookmark the permalink. 2 Comments.

  1. Hi Nikola.
    Good post…
    I thing the main thing in the last year is that MS has stopped to play a 'blind watcher' of dev community. I appreciate this on VS 2010 mainly…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: