Bringing Smiles to the Faces of MacOS developers - Miguel de Icaza
tirania.org
tirania.org
What I am peddling here is FUD, but it's warranted FUD, there are too many unknowns for what amounts to a buggy C#. Isn't the C# selling point that it's a better Java? How does that value proposition hold up for a B-grade C# with patchy coverage of the forms et cetera, with a patchy install base and an extremely uncertain future. Mono has totally failed to gain a critical mass, so there will be no openoffice.org=>libreoffice. When the takeover of Novell comes it will die a horrible and withering death getting further and further behind the parity with the MS implementation which it has never yet matched. I am sorry to say it, I wish the legal situation and Novell situation were otherwise and people could just choose on the basis of technical merits: but that's not the case, and anyone basing a startup or business on anything to do with mono is certifiably insane.
Yes Mono's future is shaky, but as a .NET developer I can adopt to Mono pretty much overnight. So if it goes belly up I haven't lost all that much.
Would I base a business on anything Mono? No, I wouldn't, you are right there. But despite all the FUD, there is value in Mono as long as you can maintain a realistic outlook on it.
If your concern are licensing the patents, MonoMac only uses the ECMA core and the MS-PL bits. If you want to use stuff outside of this and you are still scared of Microsoft going after your very profitable enterprise, remember that Microsoft is open to licensing their patent portfolio to anyone that wants it.
If Mono were to be discontinued, there is a business right there for an enterpreneur: negotiate with Microsoft a license deal for their patents, add a markup, pocket the difference.
That is the worst case scenario, you have to pay the patent holders.
That being said, you do live in a dark, sad and terrifying place. You harvest what you plant, and this fear and horror that you plant can not be good for anyone's health.
"Can" but who will? You already implicitly concede the point that mono has no community strong enough to survive without the Novell employees. Your hypothetical entrepreneur that fills this gap is charging money. Hasn't that just incurred an extra cost on the person who built their business on a free Mono? That's if this guy turns up on time for the crucial expansion to your business; if he turns up at all. There may well not be enough users of Mono to support such a model: I can't name a single big or medium sized group using Mono besides Novell itself.
Edit:
I think you have conceded my point against using Mono for a startup i.e. there is a real chance you are going to incur future costs of an unknown size. I don't live in a dark place either, please drop that language, I said I wish it wasn't this way. It's just that, given there are plenty of other options, why pick the horse lagging behind with vultures circling it for your business?
Edit2:
Also, to score a cheap point, I will highlight your use of language "fear and horror". Not only am I planting garden variety fear, but also horror. A new, and more terrifying, breed of FUD. FHUD - Fear, Horror, Uncertainty and Doubt. It's catchy.
I think you have conceded my point against using Mono
for a startup i.e. there is a real chance you are going
to incur future costs of an unknown size.
Think again ... at least one of the Oracle patents (IANAL, but have been reading the independent claims) applies to any VM that starts interpreting code before doing compiling, which applies to many tracing compilers (from my understanding, LuaJIT, JavascriptCore are good examples).
You think you're safe, think again.
What about Unity (http://unity3d.com/) and all the games based on unity. I'd say that is at least a medium size group.
Edit, links to relevant information Unity's website:
* List of some of the games made with Unity: http://unity3d.com/gallery/game-list/
* Where it says that unity is Mono based: http://unity3d.com/unity/engine/programming
For example, I've taken a semi-detailed look at both Java and C#. I find C# and .Net to be very pleasant to work in. I would totally choose C#/.Net for my own startup over Java.
You're wrong. It would be pretty much impossible for Microsoft to sue over Mono at this point, even if they wanted to. You gloss over the fact that Microsoft has issued a thorough, unrevokable, legally binding Community Promise to not assert patent claims over .NET implementations:
http://www.microsoft.com/interop/cp/default.mspx
The purpose and implications of the Community Promise are clear. Microsoft has sanctioned Mono, and no court in the world would let them take the CP back and say "nah, we were just kidding."
> What I am peddling here is FUD, but it's warranted FUD, there are too many unknowns for what amounts to a buggy C#.
Oh please. Ignoring the fact that you appear to be confusing C# and the CLR here, do you actually have any concrete complaints about mcs, or is this just more unwarranted FUD?
Yes, the central point of my comment that you didn't address.[1] That is, Novell looks like it's going to be acquired and there seems to be no reason for the purchaser to continue with Mono. It's a totally uncertain future for Mono. As I said elsewhere I am not slandering Mono, I wish it were otherwise but, given that there are just so many other cool technologies out there, what's the compelling case to bet your business on it with such a rain cloud over it? So it's not about using it for some little side project, its about founding your startup idea with Mono at the core of the technology stack, are you going to take the risk? Would you, personally, use it over the other options for your business when it doesn't have an independent community, only a soon-to-be-acquired Novell?
Listen to the squawking and preening going on in this blog to display value to Oracle. http://www.novell.com/prblogs/?p=3070
Do you trust Oracle with your business future, given the multiple fiascos that have come out of the SUN takeover?
Contrary to other "open-source" projects, Mono has a very active (albeit small) community behind it. Right now there are very active contributors that aren't working for Novell.
Also, making a parallel with another active project, JRuby, its leader quit Sun/Oracle but it is still working on JRuby ... and I've been watching Miguel's commitment and enthusiasm about Mono ever since the start, and it has only grown since then.
It's a totally uncertain future for Mono.
The patents grant by Microsoft also includes supersets (contrary to Java SE implementations), and the ECMA standard you have to implement is rather small, also making it suitable for mobile phones (MonoTouch falls under that community promise, even though it ships with a fraction of what is known as .NET)
It's also an open-source project that doesn't fall under the GPL trap ... yeah, I said it. The class libraries are MIT X11, the runtime is LGPL, it is appropriate for embedding in commercial deployments right now.
If commercial support is dropped, it can live on.
Get over it!
Microsoft is an entity independent of the people who currently work there. It's not about MS going to court and saying "We had our fingers crossed", it's about new management coming in and claiming previous management did something crazy. Do this promise protect against that?
I remember when Microsoft claimed to open their Office document formats. I argued then that I didn't know how (legal loophole, impossible to implement, etc) -- but there was a catch. A monopoly company don't want to do that -- "follow the money".
Later, it was found out that the specification was thousands of pages long, etc.
Are there e.g. any creative loopholes in that agreement? How do you know? Has that Promise been analyzed by a good team of lawyers?
Has Mono been checked for other patent troubles? E.g. doesn't Microsoft itself pay licenses to Oracle, because .NET is too similar to the Java environment?
Edit: I do hope I'm wrong, of course. (Fixed a bit of grammar.)
FUD is FUD, intelligent people make technology decisions based on facts and costs, not fear.
How does that value proposition hold up for a B-grade
C# with patchy coverage of the forms et cetera
The generational GC took a long time to get developed, yes, but at least it does have a beta quality generational GC that you can use right now (compared to the alternatives) + Mono 2.8 also ships with a beta-quality JITer based on LLVM.
And about B-grade Windows Forms ... because you know, Swing is so much better than just porting the UI layer to WinForms / GTK# / MonoMac (Cocoa) and providing a native experience.
I open the new file dialog. Type in a name, try to copy the name... and it closes the new file dialog! Apparently Cmd-C is bound to cancel. Okay, so it's a little quirky. Let's try again.
I create a new file. Type some text. Select Edit->Copy this time and... it's crashed.
Maybe I'm unlucky.
I install MonoMac, restart, create a project, write some code to display an NSAlert, press build and
Build failed. ~/.config/MonoDevelop/addins/MonoDevelop.MonoMac.2.4.0.5/mono-version-check does not exist.
EDIT: As a bonus, a process called mono has been running at 100% CPU for the last half an hour.
EDIT 2: After using it for a bit longer, the editor itself is actually not bad. It has all the features you'd expect and autocomplete seems well implemented. But this software is just far too buggy for any kind of serious day-to-day use.
Really? Tears of pain?
Just curios, because if it is anything remotely similar to selling commercial products or closed web-services (as most developers here do), you can't talk about trapping users or about the virtues of free-software with a straight face: that's called a double standard.
And most people that do that (according to surveys I've seen) are either still in highschool/college or just hypocrites ... with the occasional "real deal" that manages to live off consulting for open-source development (core project devs mostly) and writing books / training.
So which is it?
To put it another way, I imagine porting a MonoMac app to Windows would involve a similar level of effort to porting a Windows Forms app to WPF, or vice versa. The more code that lives outside of the presentation layer itself, the easier. To my knowledge, code written for MonoMac will be very Mac-specific (you'll have lots of NS* objects in there), and so it's not a matter of flipping a compiler switch to get a GTK# app or something similar. You'll have to re-write those particular sections.
Again, though, if you architect your project well, you can minimize the amount of re-writing that has to be done.
Our recommendation to developers is that they split their applications across the User Experience/Business Logic fault line. Developers then can replace the user experience with native user interfaces for each platform.
This is a bit abstract, so we tend to tell people to build a lot of UI-less reusable libraries and for libraries that have UIs, to split their Model-View-Controller libraries in a way that the Model and its support infrastructure are reusable and the View and the Controller are UI-specific.
This extends a little bit further. You also might end up with an "OS" layer that performs certain tasks in a different way depending on the operating system that you will be running.
For instance, while running on an iPhone, you have access to the gyroscope, something that you wont find on say a Linux desktop. Or on Linux, you will have access to /proc, which you wont have access to on Windows Mobile 7.
A nice example was Paint.NET. When that application was built, the operating-system specific features were all placed on a library called SystemLayer.dll. The UI was all Windows.Forms, but porting the program turned out to be limited to replacing the SystemLayer.dll with one that spoke Posix as opposed to speaking Win32.