.NET Culture Shock
bloritsch.d-haven.net
bloritsch.d-haven.net
I was a .NET consultant since Beta 2 shipped (sometime in 2000 iirc). Before that I was doing a lot of hardcore Win32 dev, mostly doing UI related stuff (was doing antialiased, alpha-channeled interfaces back on Win95 in ASM & Pascal). Long story short, I was pretty deeply entrenched in the MSFT development camp.
But once .NET hit, I was already well into a career doing web development at the dotcom factories in NYC (razorfish, RGA, Rare Medium - plus consulting for FCB/FuelNA, McCann, other agencies). .NET quickly became the platform of choice for a variety of reasons, but looking back I think it was mostly because it was easy and you didn't have to be a rocket scientist to get results. MSFT is fantastic at documentation and the supporting ecosphere for .NET folks was pretty impressive. Gone were all the XML config files, the design pattern saturation of Java, and one of the best* IDE's available for lazy, non-challenged developers, aka Visual Studio.
And it was easy money. I was charging upwards of 250/hr consulting for Pfizer doing the most inane shit you could possibly imagine, shit you could do with two eyes closed and tripping on mescaline. But people were amazed at how quickly you got shit done, but not necessarily because you were some kind of rockstar - it was just easy.
But at some point it grew tiresome. ASP.NET is a hulking piece of shit, anyone that tells you otherwise is lying and/or too stupid to realize it. You start realizing that these other .NET developers you know are unchallenged nitwits without any sense of vision, any sense of craft. A bunch of me too dim wits who collect paychecks by copying and pasting their way to completion.
I was asked about 4 or 5 years ago to take on the CTO role at http://massify.com/. I was initially brought in to consult on getting it built. My first spec and budget was insane because I was relying on the MS crutch. I looked at their goals and what it would take to run this on .NET and the two didn't align. So I stepped back and redid the spec/budget based on OSS. The cost savings were astronomical, somewhere in the neighborhood of 900%. The only problem is that the only Un*x-like system I had ever touched were the NeXT stations from college (fo shizzle).
Considering that I'm entirely self taught (I went to Art School) and the way I had taught myself was throwing myself into the fire, I figured I could figure it out and make it work. And I did. I made a few mistakes of course, but as of today - I am not longer at Massify - they are still running with no unexpected downtime for roughly ~$500 a month.
And now that I've lived in the OSS/Unix world for the last 5 years, I will never ever ever go back to Windows or .NET (though I'm the CTO for a shop whose system was written in .NET - we are currently porting it over to OSS stack). It's so much better over on this side of things, I can't even enumerate the ways.
First of all is cost savings. As a CTO, that's your primary gig (usually). When I started Massify, hosting alone was ~6K a month. After a few months of hard work, we got it down to $500 a month by pushing it into the cloud. This includes a self-scaling application that does HD media conversion, CDN hosting and distribution, the whole nine. Not a trivial application by any stretch of the imagination.
Second of all is ease of administration. The learning curve was steep, don't get me wrong, but fuck all if I'll ever go back to Remote Desktop and the bullshit that is Win2K server admin. Life is too short. I have SSHd and PowerShell rocking now for the .NET app we are currently maintaining/porting, but it's such a sorry knock off of the real thing. Gimme bash or zsh or give me death is my new motto.
Third of all is the pace of technology. Nobody is innovating in the .NET space except for MS. Everyone else is a me too, redmond regurgitating drone army. OSS moves much faster. A problem you had a month ago was probably just solved today. And you didn't have to wait for MS to do it and you don't have to pay someone bizarre licensing fees for it.
Finally, the culture is so much better. People share, people contribute, there is a collectivism that is lacking from the .NET world where the old shareware model is the status quo.
I'm not sure how MS comes back to the fore, I think it's too late for them. The world is open now, and the closed model only flies for people without a clue. To become dominant again, they'll have to open up, truly open up. But, I think it's too late. There is better stuff, for free, out there and without an OSS model, they'll never be able to catch up.
Just my .02.
You're talking before the fairly meteoric rise in the standard of web programming over the past few years. Before blogs. When tables vs divs was an actual argument. When C++ was relevant in anything apart from high performance programming.
You're also talking about .Net 2.0, pre-LINQ, extension methods and MVC.
And reading your post makes it sound more that you have a chip on your shoulder about bad programmers more than anything else:
lazy, non-challenged developers
ASP.NET is a hulking piece of shit, anyone that tells you otherwise is lying and/or too stupid to realize it
other .NET developers you know are unchallenged nitwits without any sense of vision, any sense of craft. A bunch of me too dim wits who collect paychecks by copying and pasting their way to completion
redmond regurgitating drone army
There are lots of jobs in .Net or java. There's sweet fa for a crap programmer in OSS. Once you realise that is the reason you came across so many crap programmers, perhaps you can understand that there's nothing actually wrong with .Net, it was actually a pretty fucking big revolution. On the other hand you will still come across .Net programmers who think datatables are pretty funky, rely heavily on view state and bizarrely believe .Net controls aren't the spawn of the devil.
As for the rest of it, honestly, 'Win2K server admin', um, it's moved on man. Stop beating the straw doll. 'The world is open now', look a little more objectively at the world we actually live in. Did you even notice the iPhone come out a few years back? Or the Kindle? Or all the other closed source proprietary systems that a me too OS movement gleefully copies, badly for the most part. There's plenty of room at the table for both.
Yes MS have turned into a pretty crap company under Ballmer, but their language teams are still pretty damn good.
The reality is you've become a more experienced programmer. And somehow you're magically attributing it to OSS instead of just knowing more than when you were wet behind the ears.
OSS and proprietary seem on opposite sides of support-systems and customer-facing systems. Proprietary > OSS for users' visual cortex, OSS > proprietary for everything supporting that top layer.
I think part of it is that OSS UI reeks of design-by-committee. Which is something Apple has gotten ridiculously correct. UI needs to be unified and have a flow to it, something which you're more likely to get from a single (highly skilled) person than from a thousand.
I hate reading this. It shows little experience with with open source programs, and ignorance of how open source programs work.
There are hardly any OSS projects with thousands of people- in fact, the only one I can think of is the Linux kernel. Most GUIs in the open source world are worked on by one or two people. Firefox is a good example. Ben Goodger and (one?) other person did essentially all the UI work.
Now, it's true that there are a lot of open source programs with terrible user interfaces. Some of this is because of the strong connection between *nix and OSS, and the primitive toolkits available 15 years ago. Some of this is because it's written without an end user focus. Sometimes the authors don't care. Some of it just sucks.
It's true that Apple has had a lot of success by paying a lot of attention to UI issues. But they're not only different from the majority of OSS software, they're different from the majority of proprietary software. Have you ever tried to explain to someone how to use Opera, or Internet Explorer? Word? Security products like Symantec? The world is filled with baffling crap, it has nothing to do with the licensing scheme.
In short, I am bothered by your generalization because it I believe it to be untrue, and worse, unexamined.
Just look at Ubuntu's interface preferences. Committee. Everyone wants an option, so they're all there. Perfect for alienating everyone who doesn't want all the options. I.e., 99% of the world. Look at <OSS program X> and poke around (especially the popular ones - try Open Office). You'll find bajillions of settings which prevent average people from going deeper and improving their experience.
edit: granted, this is changing in a few areas. Rails, for instance, is convention over configuration - and look at how successful it's been. Mobile apps have been revolutionized by Apple's apps, and developers all over are rushing to mimic it. People are starting to learn, but it's a long way to the top, and the vast majority of the improvements seem to be originating from companies, not OSS groups.
OpenOffice.org is a disaster whose problems go far deeper than the UI.
Also, rails is changing things by preferring convention over configuration? What year is this, 2005?
In the end, here is my complaint: You are taking what is essentially the best case scenario, proprietary or not, and saying that it means software with a copyright license is more usable than software with a copyleft license. Stop doing that, it does the general discourse about software a disservice.
I've yet to see more than low-double-digits of beautifully designed open source programs, and the majority of those were made entirely by one person, often open-sourced because they didn't think they could get enough money to make monetizing it worthwhile. I've been watching for years (though I could be watching in the wrong area). Meanwhile, the rest of the world has been utterly rocketing towards more focus on design, and has a long history of success.
Want more evidence? Look at games. Millions upon millions of players, probably approaching billions of copies sold / downloaded, very few of which are open source games. There's an entire industry oriented around UI/UX and design... and where's the open source influence? Bad Civilization / Quake clones? A lot of the unique designs I've seen have been done by individuals without open sourcing the code, for whatever reason.
And yes, OOo is frightening on many levels. It's a worst case scenario, and I recognize it... but I see it echoed frequently in other OSS projects.
I have not said OSS cannot produce good UIs. I have said there seems to be a trend, and at least I have seen far less. And also take into account that this is not done by enumerating all programs ever produced; this is weighted by success, especially with the design-oriented consumer culture lately.
Try comparing native/native or virtualized/virtualized. The vmware and virtualbox drivers lack many functions that are used in native-running ubuntu or windows, for that matter.
I hate trolls like you. A lot of us work really hard on making it the best damn platform ever, and it is, and then some quack who doesn't know anything says "it's just like Windows 95" which shows that they didn't even bother to try it out, or take the tour on ubuntu.com or even watch a recent video of the latest release on youtube. And that just sucks the fun out of life.
Please, actually take a look at it before saying something ridiculous.
Calling users trolls who don't know anything doesn't make me want to give you any credit. Vetinari made me consider that I might be missing out on the best the platform has to offer but you just made me want to uninstall it.
Better by far, certainly. The Ubuntu crowd has made significant strides in the past few years. But just look at the "Ubuntu One Music Store" image and tell me you don't see glaring inconsistencies in padding, form outlines, and control styles. Then look at other applications and notice which buttons are shaded and when, and resizable edge styles, and why does the shade get that dark near the bottom without a font highlight to counter the harder-to-read lower contrast?
Wow! I was building sites with Zope by 2002. There were far better alternatives than C++, Java and .NET far earlier than 5 years ago. I did lots and lots of Perl since 1996 and some PHP thrown in around 2000. Only used C++ and Java when nothing else would do (mostly because of clients that had too much of the wrong Kool-Aid).
Microsoft has been crappy for a very long time.
And yes, Microsoft has been trying to get more and more Unixy over the years, and more and more Appley, and I'll bet they continue doing that in the future. Good for them.
About what domain are you talking? Perl and PHP were also quite popular in web development as far as I can recall. And several others.
It lacks support for something as simple as automatically inserting the method declarations for interfaces when you had to implement them, let alone promotion of variable declaration to fields. And there is essentially no way to uninline a method. Eclipse has _all_ these things.
And this is still a problem for the 2010 ultimate edition.
He says that .Net coders don't answer each other's questions. What's stackoverflow then?
He says that the .Net ecosystem is not as rich, and mentions nUnit, but not the other major open-source testing frameworks - xUnit and mbUnit. Then there's also MS testing, or if you like BDD, there's NSpec and SpecFlow.
I've seen seven or eight open-source .Net IoC containers (including one from Microsoft). By far the weakest of the major ones was Spring.Net, the straight-up port of the popular Java tool. If you don't like EF (Microsoft's ORM), there's NHibernate. Subsonic and others. If you don't like Webforms or asp.Net MVC, theres Fubu and OpenRasta.
Unfortunately a Resharper licence is the price of entry into a lot of IDE goodness. He has that right anyway.
The .Net shop that I work in doesn't choose tools to use in our .Net development efforts based on cost or licence. We choose what makes sense. There is really are few or none "only paid, proprietary software" development shops left. It's all mixed.
The first response in the comments secition is plug for stack overflow. The author didn't know that it existed.
I find it quite telling. Every company I know of that blocks unauthorized websites, IRC channels and so on happens to be .NET (or other flavors of Microsoft) shops.
But the reverse is not true - not every company on .Net blocks websites. I don't actually know of any that block.
Blocking digg or facebook is one thing, but If you block Stackoverflow, there's no hope for your development practices.
It's not without its faults, but a lot of the negativity expressed towards it is unwarranted.
Try Haskell, Lisp, Forth, assembly (on bare metal - write your own kernel), Erlang and anything that's "too weird" for most "enteprisey" developers. Buy an old 8-bit computer at eBay just to get a hang of what is programming within 64K of memory. Make hardware for it. Play with non-relational databases.
Most important advice of all: play. You are not learning a job, you are becoming a craftsman (craftsperson?). If you don't enjoy it, you won't do it well.
You'll have time to learn Java, .NET and even Python and Ruby later.
I did the same thing (Lisp, Forth, BSD, OCaml, playing with an old pizzabox Sparcstation), but I was a history major. :)
Sure, in college, you should work with more academic and esoteric languages. I learned Lisp in college.
The point is, many people dismiss the merits of Microsoft tools and languages. One guy I worked with said, "I haven't used any Microsoft products in years" and that's supposed to make you an expert on them somehow? :)
No. But it has a good chance of making their experience broader. Microsoft is nothing but a tiny spec in the software development tools set. Using only Microsoft is not a good sign.
On the other hand, perfectly good things like C# (especially with LINQ and ASP.NET MVC) are being dismissed out-of-hand by otherwise reasonable people just because they came from Redmond.
I agree, C# is a better language than Java, but that doesn't mean I'm going to use it. There's more to choosing a language than the language itself. The environment is at least as important.
I wonder if there's any chance of MS ever porting CLR to the *nixes...
How did you found out about it?
I can't find a mailing list. Do you know of some way of contacting the author?
IMHO .NET still has one huge advantage - developing UIs for enterprise apps. Very easy to learn drag'n'drop win/web forms + multiple control vendors. Real component oriented programming.
A lot of what he says is true (developers who use Microsoft products are generally more willing to pay for software). but there are plenty of open source projects and a TON of online references - and there were even before the latest and greatest stack overflow came into existance.
"never heard of stackoverflow" is much the same as "never googled a tricky .Net question". Someone's due for a dose of culture-shock.
Here's the main distinction: Whereas Java has underlying it more Java, .NET has underlying it 30 years of carefully-optimised Windows C/C++ code. It's basically a layer turning Java-like code into very fast Windows code.
And it's clear to me that Microsoft spent a lot of money writing this layer over Windows that makes writing Windows programs so easy. People don't give them credit for that.
You know it's not more a compliment to .NET than a harsh critique on the (utterly detestable) Win32 API.
> Whereas Java has underlying it more Java, .NET has underlying it 30 years of carefully-optimised Windows C/C++ code
Your faith in humanity impresses me.
So I've seen this optimisation with my own eyes.
I also believe the Java stack has some very carefully optimized code too. It certainly has the performance to back up that affirmation. Besides that, it was designed to be easily portable to other OSs and architectures, a feature that's more important, IMHO, than speed under Windows.
Finally, you should only optimize for speed the code your profiler says has big impact on your performance. The rest you should optimize for readability and correctness. There is a lot of very subtle bugs we introduce when trying to optimize stuff and it's much easier to optimize correct code than to fix blazingly fast, heavily optimized buggy code.
Do you tell people that "an elephant is like a tree", much?
I think you are talking about WinForms there. There's a lot more to .Net than targeting the "30 years of Windows C/C++ code" that makes up WinForms.
There's WPF on the desktop, and WPF's little brother, silverlight. The's serving web pages via ASP.Net WebForms and MVC variants, And serving data over web services. One could spend several careers using .Net without using as a wrapper for Windows' C Dlls.
(with apologies to Zandorg and Douglas Adams)
Is it' Java all the way down? No. All virtual machines and runtimes are wrappers over the OS's abilities. How is Java different here?
On the flip-side are a number of robust open source .NET components that are free like NUnit, SubSonic, NHibernate, RestSharp, and others. The most impressive is probably the Mono project, which attempts to port the entire CLR stack to POSIX platforms. IronRuby and IronPython are also impressive.
Unfortunately it's missing a web stack; ASP.NET sucks on top of Mono.
So yeah, you can run ASP.NET MVC on top of Mono, but it will leak memory, the server requiring periodic restarts.
When people say "you can run ASP.NET on top of Mono" they never have an example of a visible website / web service that does that.
Also, doing asynchronous I/O on top of Mono using the model available in System.Net.Sockets is flawed on Linux, because that paradigm is based on true async sockets available on Windows ... but on Linux it's emulated with ThreadPools.
A good and scalable server for Mono would be one using async paradigms available on Linux (like bindings to libevent or other libraries, like something that calls vmsplice/splice to send/receive data). It would also have to be throughly optimized/profiled such that it relies as much as possible on stack allocations and to be careful about heap-allocation patterns, such that it doesn't leak through memory fragmentation.
Also on speed ... using ASP.NET MVC you can currently handle 500 requests/second for a simple "hello world". This is simply unacceptable ... using Java you can easily reach 3000 r/s, and this by using a regular servlet-compliant synchronous server (by using stuff built on top of NIO, you can reach 15000 r/s). It is unacceptable because ... what's the point in using a compiled language with a smaller community, other than getting raw performance out of it?
Again, this could be solved with libraries designed for Mono. But unfortunately not too many people are interested in Mono as a web development platform.
There's nothing wrong with charging for writing code. It's when you charge people to use your code (that they will need to make modifications to to get it to work) that it becomes detrimental to the community.
On the other hand, it's less of an issue if you're charging people for something that works out of the box. That said, in my experience just about nothing works out of the box.
The incentives in the Java world are completely wrong. It's best to make your library as complicated as possible, because then you'll get a lot of people buying support. If you charge for the code then you make it as easy to use as possible to avoid support.
Counterpoint: You know and can use the code before you pay anything. If you decide to prefer a certain library over another, this decision is based on real-world experience with that library. This decision might imply that you pay for support later on -- where support sometimes means implementing requested features.
Fortunately the way humans are wired makes it relatively easy to get out of this trap: feature lists. Humans love long feature lists. You can add a lot of complexity without scaring people off.
> you're implying that the incentives are wrong with OSS in general
Yes, I am implying that for code that isn't developed for use by the company itself, or by non commercial programmers. This is very unfortunate, but whether you like OSS does not (or should not) change this. I have been thinking about ways of avoiding this, but so far I have not found a model. What I want is a business model where:
1. The programmers do what they love and do best: coding, not support
2. The primary activity (programming) brings money in, not some secondary activity
3. Good libraries should earn more money
4. The code is open source
A dual licensing model like Qt seems like the only way.
Sadly the management seems to be considering that all new development should be in .Net and centralised with TFS. I can't rightly understand why anyone would want to tie themselves to this costly, closed platform when they can essentially choose anything.
I've seen .Net code stored in TFS, svn, perforce, Mercurial and git. If your management aren't using the source control system, why are they deciding? Just use the one that suits you.
It's an interesting take on .NET he has. I find that answers and sharing is as prominent in the .NET community as the Java community -- perhaps not as many F/OSS projects. Dunnno.
As far as setting up NUnit, takes about 5 minutes. Sure, it'd be better if there was a whizbang plugin for the IDE, but if you wanted you could make one easily enough and distribute. It's not convenient, but it never seemed like a huge deal. I've used NUnit, XUnit, and the MS tools. If I remember correctly, seems like there is an open-source pluggable test running tool. Never have played around with it.
And if you need help? Google is your friend. There are so many .NET people out there blogging and sharing code that you can just about ask Google any problem you might have. Hell, I've copied the error right from the IDE and pasted it in the search box -- works almost every time.
I like both communities, although I have become less of a fan of MS over the years. Lately I'm leaning towards coding in F# on mono for new projects, but I haven't set up the stack yet.
He does have a point, there is a difference in the cultures. But both communities are very accessible and easy to work with, at least from my experience.
Though--be careful with the term "real programmer".
My friend Johnathan calls the others "career programmers". Those are the people that you end up baby-handling, qualifying your statements and padding your speech with descriptive wiki paragraphs lest you lose them to a fancy day-dream.
"A particular sub-variety of hacker: one possessed of a flippant attitude toward complexity that is arrogant even when justified by experience. The archetypal Real Programmer likes to program on the bare metal and is very good at same, remembers the binary opcodes for every machine he has ever programmed, thinks that HLLs are sissy, and uses a debugger to edit his code because full-screen editors are for wimps. Real Programmers aren't satisfied with code that hasn't been tuned into a state of tenseness just short of rupture. Real Programmers never use comments or write documentation: “If it was hard to write”, says the Real Programmer, “it should be hard to understand.” Real Programmers can make machines do things that were never in their spec sheets; in fact, they are seldom really happy unless doing so. A Real Programmer's code can awe with its fiendish brilliance, even as its crockishness appalls. Real Programmers live on junk food and coffee, hang line-printer art on their walls, and terrify the crap out of other programmers — because someday, somebody else might have to try to understand their code in order to change it. Their successors generally consider it a Good Thing that there aren't many Real Programmers around any more. For a famous (and somewhat more positive) portrait of a Real Programmer, see The Story of Mel' in Appendix A. The term itself was popularized by a letter to the editor in the July 1983 Datamation titled Real Programmers Don't Use Pascal by Ed Post, still circulating on Usenet and Internet in on-line form." (http://www.catb.org/jargon/html/R/Real-Programmer.html)