A Bare-Bones Linux+Mono+GUI Distro?

nimble99 writes "I am a computer software engineer, focused mainly on the Windows platform — but most of my development time is spent in .NET. I would like to move my .NET development to Linux in the form of Mono, in an attempt at building a media-center type of device. All I require, is a base operating system with simple hardware support, Mono, and a window manager that (preferably) does nothing but act as a host for mono applications. Is this available? I dont know a lot about Linux, so I thought I would ask if there is already something like this available. Obviously a 'Mono Operating System' would be the cleanest solution, but a similar thing could be achieved with the barest minimum of Linux distros right?"

Gentoo

[armanox]

You could build a Gentoo install to satisfy this. With Gentoo you build the system to fit what you want out of it.

Re:

[pm_rat_poison]

Another similar choice would be Linux from Scratch. Choose whichever one you like. They can be quite a hassle to set up, but the reward is that you have a completely customized system, with complete control on what's installed.

Re:

[nurb432]

Or perhaps a minimal install of Freebsd + mono

I know he said linux in the question, but installing a minimal Freebsd box is much easier. And if he mostly cares about mono, does it really matter what OS its running on?

Monoppix?

[dreemernj]

Monoppix perhaps? [monoppix.com]

I haven't tried it yet but the description sounds about right.

Re:

[Otter]

That's an interesting link but it doesn't sound like the minimalist setup he wants.

Doing it himself seems like the best bet. Certainly if he's planning to develop a commercial product based on this, it's worth figuring out a custom distro that does exactly what he wants.

Re:

[RiotingPacifist]

given that he's new to linux, he may consider that a year down the line but for now monoppix is his best bet

Re:

[Otter]

I understood his question differently than most of you did: I thought he was asking about a platform for the device he plans to build, not just a personal development platform.

If the latter is correct, I'd suggest whatever distro his friends or coworkers can best help get started with (Ubuntu, Fedora, Gentoo, whatever) and not worry about Mono-specific wrinkles.

Re:

[hey!]

Well, I think you make a really good point.

When you are developing, you are a user. Which means you are a pain in the ass to somebody.

Once the box goes out the door, it means it's in the hands of users, who will be a pain in the ass to you.

The way to make money in tech is to find something that people will pay money for (duh), but the important point is that it doen't have to be much money, so long as they never ever call you for support. I know a guy who launched possibly as many as a dozen commercial

Ubuntu Server Edition

[Ngarrang]

Ubuntu Server edition installs as just a command line without all the fluff. From there you could add Mono and any GUI of your choice.

Re:

[bsDaemon]

Without all the fluff, how is it not just Debian?

Re:

[Ngarrang]

It has all of Ubuntu updatedness in the kernal that might not be present in the general Debian release? I have no freakin' idea, actually. The majority of my linux experience is Ubuntu. I have a learner box that is just running Server edition on which I have installed and uninstalled different GUI's and been learning the command line.

I do know the server edition is tuned to give background tasks more priority.

Re:

[croddy]

ubuntu's APT repositories are separate from debian's, and many packages are at different versions due to the way in which ubuntu is periodically synchronized with sid and then bugfixed by ubuntu people. ubuntu's metapackages also differ from those found in debian.

Re:

[nhaines]

Because Ubuntu releases an update every six months and Debian stable (this is a server we're talking about) releases an update every two years.

Debian and Ubuntu maintain the same version of their packages as is released for each version, generally applying only bugfixes.

Re:

[MasterOfMagic]

This one has screaming Linux fangirls and fanboys throwing their panties and bras on stage at your feet. And something about going to 11. Petrified.

Re:

[nuzak]

Ubuntu iterates through stable versions much faster than Debian -- even the "long term support" versions come out faster than Debian Stable.

I run Sid myself, but you can't get your security updates separate from your "might break all your dependencies" functionality updates. For my single development box, that's fine, but when you've got multiple production deployments, that's really bad.

Debian or Ubuntu Minimal.

[SanityInAnarchy]

Just grab the Ubuntu Alternate install CD, or start from Debian. Same principle, but no real point in getting Ubuntu Server -- you probably want the generic kernel anyway, not the server kernel.

Debian

[Hatta]

Just install a basic net installation of Debian. You'll get nothing but a console. Apt-get the GUI of your choice. Apt-get Mono. You're done.

Damn Small Linux

[pavon]

DSL is also a good choice for a set-top box - only 50MB of disk space required, and it is based on Debian so pretty much any application you want/need is already packaged and available in the repository.

That said, if you are going to be doing Mono development you will want a full desktop environment - for that any distro will work fine. I'd go with a full Debian install on your development machine so you have the same library versions/builds on both systems.

For the final media center PC, you don't actually

OpenSuse Vmware Image.

[megalex]

If you are on windows and have decent ram you can try the mono vmware image. It boots opensuse desktop and has mono and monodevelop ready to go. go here: http://www.go-mono.com/mono-downloads/download.html [go-mono.com] Click on the vmware image.

Re:

[somersault]

Anyone care to explain why this is '+2 funny'? It seems a pretty useful idea for the initial development stages at least.

Re:

[Alex Belits]

VMWare emulates a fixed (and $deity-awful) set of hardware, it's completely unusable for any development targeted at embedded platform.

Re:

[somersault]

Ah. It still makes sense from a 'get-rid-of-the-licensing-cost' point of view, and as an initial testing environment without having to reboot or use separate hardware ..

Try SuSE

[Ed Avis]

Mono is developed by Novell, so if that's your main app it would make sense to use their Linux distribution, SuSE. Either as OpenSuSE (or whatever the capitalization is these days; cf NeXTStep) or SLES. It is not minimal but it includes the latest Mono stuff and you can probably pay for support if you want. Since there is some overlap between Mono developers and GNOME developers and some GNOME applications like Banshee, F-Spot and Tomboy are written in C#, it probably makes sense to use GNOME as your desktop environment.

That said, I'm quite happy with Fedora, Mono packages are included, and if you need something more recent than the last Fedora version you can easily compile it yourself.

Your job is to be a software developer, not a desktop-customization weenie. So forget about spending time on making or finding a 'minimal' environment. Any modern Linux distribution won't get in your way and will let you get on with porting your apps to Mono.

No you don't

[LWATCDR]

No you probably don't.
What about a web browser for research? An IRC client for asking questions?
Ubuntu is will do what you want. It may have a few extras you don't need so uninstall them.

Qt...

[ninevoltz]

Use Qt4 and forget about MONO and .NET. You won't regret it.

Re:

[nawcom]

agreed.

Re:

[pembo13]

I like everything I've read about Qt4. But you can use Qt4 with .NET

Re:

[Alex Belits]

You can also bake cookies on a space station, however that would be neither an efficient way to bake cookies, nor a good use for a space station.

Re:

[vossman77]

That is the greatest quote I've heard in a while. I'm putting it up in the cubicle.

Re:

[losinggeneration]

Or maybe they could check out Qyoto [qyoto.org]

Re:

[cbrocious]

Oh, so Qt4 allows for complex runtime code generation? It allows you to deploy a single binary on multiple platforms? You're not just comparing comparing two completely different things here, you're comparing two completely different classes of things.

Re:

[qbast]

Yeah, it allows to use single binary on multiple platforms as much as .NET: you just compile it for windows and run it with wine on non-Windows. I don't see much difference between win32 app running on 'alternative' win32 API implementation (wine) and .NET app running on 'alternative' .NET implementation (mono). Well, except for the fact that wine is more mature and more likely to work properly. Miguel may argue otherwise, but all we got from mono is another windows subset emulator doomed to always play cat

OpenSuse Mono Live CD

[megalex]

You can also try the live CD for opensuse its available in the download section http://www.go-mono.com/mono-downloads/download.html [go-mono.com]

Look at Debian Live

[SeanTobin]

I've been playing around with Debian Live [debian.org] recently and the level of control you have over package selection and customization is impressive. It takes a little work to get used to the build system and how to customize your final image, but after you get through it once it is very simple.

You also have the ability to build images for CD-Rom's, usb sticks, netboot or hard drive images.

If you are not familiar with Linux, this route may be like jumping into the deep end. As others have mentioned, you may be better off using a canned distro like Monoppix [monoppix.com] while you do your development so you know exactly what you need in the end. Once you are comfortable and ready to move toward your final product, look towards Debian live.

Warning: The following contains stupid questions.

[mckinleyn]

There are an infinite number of solutions to this problem. Everything from Gentoo to Knoppix to Puppy Linux (for the love of God, avoid the last).

Do you:

( ) Intend to use it as a liveCD?
( ) Want disk tools in the installer to allow you to keep a windows partition?
( ) Want to avoid significant configuration?
( ) Have no Linux experience whatsoever?
( ) Want it to exist in a partition WITH Windows?

( ) Have better things to do than sift through the inevitable 2541 comments?

Th

DSL + modification

[nicholasperez]

I had this exact requirement for building my point of sale device. My solution was to go with something like Damn Small Linux [damnsmalllinux.org] and modifying it. Basically, I altered it to be apt enabled yet again instead of using their package management and then pointed it at the debian repos. One apt-get install later and I had everything I needed to host a .NET GUI application. You could probably get away with a minimal debian install as well, but DSL and its brethren generally have good package selection for smallness.

Try windowmaker

[poeidon1]

its the simplest and lightest windowmanager that I have seen. Ofcourse, tthere are probably other options. Generally, the packages are available for most distros but I would advise you to use gentoo since you can make it as lightweight as you want depending on your requirements, but gentoo requires you to know a thing or two already about linux.

OK, I'll bite...

[Noryungi]

I don't know why, but this question sounds really, really weird to me: if you are doing .NET development, why move to Mono and Linux? Why not just stay under Windows, especially since you say (and I quote):

I dont know a lot about Linux, so I thought I would ask if there is already something like this available.

Anyway, here is my suggestion, but, as another poster has already pointed out, any Linux/GUI permutation would probably work just as well:

1. Slackware [slackware.com] for trhe Linux distribution.
2. Fluxbox [sourceforge.net] for the GUI.
3. Then, you can use either the Gnome Slackbuild [gnomeslackbuild.org] or the Slackbuilds [slackbuilds.org] files to compile Mono on your Slackware machine.

Some people would argue that using Slackware for this is crazy, but (a) Slackware is a lean and mean developement platform, and a very lean Linux distribution and (b) it will teach you a lot of things about Linux, and UNIX in general.

I hope this helps!

Re:

[incripshin]

I tend to agree. I started with Slackware, and it taught me most everything I know. Wait a minute ... don't use Slackware. (Just kidding on that last part.) If you were more experienced, I would recommend Gentoo to roll your own distro, since it's easy to just install only what you need. Besides, the install process was pretty fun. Too bad I never had to reinstall it in the last three years (I know, right?). I remember trying to install a links+ package in Ubuntu without an internet connection and it

Re:

[akadruid]

I don't know why, but this question sounds really, really weird to me: if you are doing .NET development, why move to Mono and Linux? Why not just stay under Windows, especially since you say (and I quote):

I dont know a lot about Linux, so I thought I would ask if there is already something like this available.

I can answer this, being in much the same postion.

Like a lot of Slashdotters, I get paid for coding, and so I use C# and .NET at work, but I love my Ubuntu setup at home. Mono lets me use a language I am comfortable in for pet projects.

Linux is perfect target environment for the kind of (ambitious) pet project he has in mind, it's far more suitable than Windows for repurposing older kit or scratch building, it has no cost or licensing hassles, drivers are built in, and everything is open.

Of course the sam

Tell us more

[Just Some Guy]

All I require, is a base operating system with simple hardware support, Mono, and a window manager that (preferably) does nothing but act as a host for mono applications. Is this available?

Is that exact arrangement pre-made? Probably not. Why don't you let us know what you're trying to accomplish so that we can steer you in the right direction?

I'm a KDE guy, but my first suggestion would be to install Ubuntu with the stock Gnome desktop. Just because you can run other applications doesn't mean that you have to.

Many Choices

[manboy9]

AFAIK, there is no such thing as a Linux-based "Mono OS". You'll probably need to modify an existing distro. Since you claim not to be a Linux expert, Ubuntu might be a good choice, but it's a bit of a heavyweight. You'd need to spend a lot of time disabling features to trim it down, which requires quite a bit of Linux knowledge, so it doesn't really help you avoid learning Linux. Same applies to Fedora, SuSE, etc. If you intend to develop on a platform that closely resembles your target platform, keep

DIY

[hlt32]

#1 Download debian netinst or ubuntu server.

#2 Install distro.

#3 aptitude search mono

#4 sudo aptitude install etc.

#5 sudo aptitude install xorg fluxbox

#6 ???

#7 Profit.

Re:

[mollymoo]

I wouldn't go for a server distribution for a multimedia GUI project without checking that it doesn't have a kernel tuned for throughput rather than latency (eg. 100Hz ticks, no preemption). Just use any old desktop distribution and turn off what's not required. It's not like a couple of gigs of disk space matters much on a box that will be storing digital video.

yeah, it's called...

[nguy]

"I am a computer software engineer, focused mainly on the Windows platform -- but most of my development time is spent in .NET. I would like to move my .NET development to Linux in the form of Mono, in an attempt at building a media-center type of device. All I require, is a base operating system with simple hardware support, Mono, and a window manager that (preferably) does nothing but act as a host for mono applications. Is this available?

Mono is not .NET. Mono is C# with a large number of bindings to FOSS, including Gtk+ and Gnome. So, that means you need a fairly complete complement of all the C libraries. If you want .NET on Linux, you need all of that, plus the .NET compatibility libraries; those are not usually installed. In addition to that, Linux needs its package management, installation, upgrade, system maintenance, indexing, and other tools. Those mean that you have to have a POSIX environment and a reasonable complement of C and C++ libraries.

So, basically, what you want is one of the basic Gnome or XFCE distributions, with the additional .NET compatibility packages installed. Ubuntu and Xubuntu are good choices.

Everybody occasionally dreams of getting rid of all the "old stuff" and just replacing it with something "modern" written entirely in the language-du-jour. But there are several reasons against that: (1) the old stuff works well enough, (2) it's not clear that you can do better, and (3) the old stuff has proven that it has staying power; C# may be gone in three years and you have to start from scratch.

I would also recommend against programming in .NET on Linux; use Gtk# and C# bindings of the Linux native libraries instead. Monodevelop should make it pretty easy to get started, and Gtk# is a reasonable and easy-to-learn toolkit.

Re:

[Scaba]

Mono is not .NET.

That must come as a surprise to the Mono developers who claim "Mono provides the necessary software to develop and run .NET client and server applications on Linux, Solaris, Mac OS X, Windows, and Unix" and "The Mono Project is an open development initiative sponsored by Novell to develop an open source, UNIX version of the Microsoft .NET development platform."

Re:

[nguy]

That must come as a surprise to the Mono developers who claim "Mono provides the necessary software to develop and run .NET client and server applications on Linux, Solaris, Mac OS X, Windows, and Unix" and "The Mono Project is an open development initiative sponsored by Novell to develop an open source, UNIX version of the Microsoft .NET development platform.

Not really. It's really not that hard to understand:

• The Mono project as a whole produces a large superset of the Microsoft .NET platform. So, the en

Re:

[HalAtWork]

Can't you just use windows.forms with Mono?

Re:

[nguy]

Can't you just use windows.forms with Mono?

You need to install the winforms libraries; they are not installed by default.

The only applications on Ubuntu that use Winforms are monodevelop (port of Sharpdevelop), IronPython (Microsoft open source software), and IKVM (JVM emulator). None of the actual mono desktop applications use winforms, hence it isn't installed by default.

And if you're developing new software for Linux with Mono, it would be foolish to use winforms: Gtk# is better and better supported (an

Re:

[nguy]

You know, I'm kind of sympathetic. I think C and C++ suck and I agree we need something to replace C and C++.

I actually kind of like the core C# language. I don't think there are patent problems. But C# is not the answer because it's too bloated, not implemented right, and changing too rapidly.

from VB to PowerShell

Those are the wrong things to say to try to convince someone to move to C# or CLR. PowerShell is another testament to Microsoft's incompetence.

Bare Linux

[treak007]

If you are looking for a barebones distro, try gentoo (gentoo.org). For a barebones window manager, perhaps try something like fluxbox or blackbox. Pretty much, gentoo will give you a base system and it will be up to you to set everything up.

Look at LFS

[ComputerSlicer23]

It's not for the feint of heart, but you might look at using LFS to build such a minimal system. I don't really see the harm in using a "full" Linux machine for the development environment, and then using LFS to build the embedded image that you deploy to "real" devices. We do this where I work.

http://www.linuxfromscratch.org/ [linuxfromscratch.org]

It'll get the job done. If your totally new to Linux, it might be a bit much, but the folks on the lists are quite helpful.

Kirby

Re:

[4e617474]

Linux From Scratch would be a good fallback position if building what you want on top of Gentoo or Arch didn't turn out to be as comparatively easy as it was supposed to be. It'll make you do a lot more work, but overcomes the snags through brute force of documentation, assuming little foreknowledge.

For a distro you can actually work in, the minimalist boot options [uni-kl.de] of Knoppix [knoppix.com] would let you do a quick restart of X and see what your apps do in something like what your final release image is like. If nothing

warning: this is slashdot!

[Goeland86]

As much as it makes sense to ask the question here, I'm afraid of the thunder of distro bashing this is going to start.
This is slashdot. We're not always rational, and we argue our tastes in linux distributions and other functionally equivalent software.

That said, someone suggested gentoo, and while as a gentoo user I applaud the suggestion, I cannot help but think that it is a little steep in the learning curve to fulfill these (relatively simple) goals. While I don't think anywhere else would like to

Xubuntu. Or stay with Windows.

[Qbertino]

Lightweight? Don't no nothin' 'bout Linux?
You want Xubuntu [xubuntu.org]. The official Ubuntu variant using the XFCE Desktop. Monodevelop will use quite a bit of GTK stuff though, IIRC.
Anyway, it's faster then Windows, so no downside here. But if Mono isn't enough, then I'd recommend staying with Windows. Unless, however, you want to learn more of Mono to slowly shake lose of MS. Monodevelop looks a very fine and dandy OSS IDE and even makes me curious about this Mono stuff.
Good luck and welcome on board.

OpenSuSE

[oliderid]

I develop in Mono. I have chosen OpenSUSE over other distibutions because it's Novell behind and I thought that Novell would be the best company to packaged correctly MonoDevelop, Mono and everything else.

The OpenSUSE installation lets you remove unwanted apps. But well I have never removed anything...Quite the contrary, You always needs specific tools.

Oh...And well you are going to develop pro applications right? So I wouldn't advice to take the last little distribution...Somes are maintained by one person mostly...If this person gets sick or is fed up by the maintenance...You are in trouble. For pro the best thing is to wait until the distribution becomes mainstream (Ubuntu, Gentoo, etc.)

Re:

[jsrlepage]

Did he ask you your opinion on the .net Framework/Stack? I am pretty much against Microsoft as much as anyone, but given the fact that their .net is pretty-much-fkinly-successful, maybe Mono can get some .net programmers to cross the bridge.

Also, Mono TECHNICALLY isn't a Windows infrastructure. It's heavily based on a Microsoft-created execution and "vm" stack (vm not so much, but i disgress), but it's a reimplementation of .net. Therefore, it is NOT a microsoft product. It's more akin to Wine than Windo

Re:

[popeyethesailor]

I think GP's comment is quite valid. His comment was on the popularity of the framework on the Linux side of the fence, which is pretty much true. Outside of Suse, I'm not sure there's a lot of traction for Mono.

Re:

[Daengbo]

Banshee, one of the really popular audio players, is based on Mono. When Banshee-1 comes out, a lot of people will be using it. Trust me.

F-Spot, the default photo app for ... ahem ... Ubuntu 8.04 is a Mono app.

Beagle is Mono.

Tomboy Notes is Mono.

The way I see it, most of the brand new, quickly developing application software is based on Mono.quickly

Re:

[bWareiWare.co.uk]

Actually it goes further then this:

Mono's .NET API implementations (WinForms etc) are to .NET what WINE is to Windows. (I.e. great for easy porting, but destined never to be perfect). Though WinForms is a much simpler and better documented API then the Win32 beast.

However Mono's VM is more comparable to x86. It is a published standard and Intel's initial development of it doesn't give them much intrinsic advantage over AMD. (And hopefully Mono's OSS nature will prove equivalent to Intel's fabrication advant

Re:

[tolan-b]

Myth is a bit of a nightmare though, and personally I don't think the media player side of it is that great.

If you don't need PVR then I can heartily recommend XBMC, it's awesome.

http://xbmc.org/ [xbmc.org]

The following contains sarcasm:

[RingDev]

Thank you for answering the question. A truly informative, well thought out, and highly insightful post. I'm sure with your great advice the asker will make great strides in his project. And following along your high standards more technical solutions and products will be introduced to the market.

Thank you,

-Rick

Re:Don't.

[halivar]

I'm a Windows developer who uses Linux at home.

As a user, I am philosophically opposed to Mono as an unwanted Windows fingerprint on my completely FOSS system (except for my nVidia drivers; alas, I am not perfect).

As a developer, I am so completely convinced of the superiority of the .NET model and the C# language (Anders Hejlsberg has been my hero ever since he developed Delphi during his Borland years) that I cannot help but prefer its usage, even in a Linux environment.

I'm thinking about developing a split personality to deal with this paradox.

Re:Don't.

[Ed Avis]

Bah. The yardstick is not 'has nothing to do with Microsoft Windows' but rather 'gives you and others freedom to use, share and change the software'. If you just wanted to eliminate Microsoft you could buy a Mac and not buy Office for it. Mono is completely free and open source software. Yes, it is a clone of a proprietary system, just as GNU started out as a clone of proprietary Unix.

Re:

[mweather]

GNU's Not Unix.

Re:

[frank_adrian314159]

I'm thinking about developing a split personality to deal with this paradox.

Both of you let us know how that's working for you!

Re:

[somersault]

I didn't realise that the guy who did Delphi had anything to do with C#. I quite like Delphi and use it for any Windows GUIs I make. I still don't really want to start using MS' proprietary anything, even if it was designed by a genius, but thanks for the info.. if MS can clean up their act with Windows 7 then I may be able to put up with them for another while, and may even look into .NET (if Delphi doesn't already use it by default).

Re:

[renoX]

>I am so completely convinced of the superiority of the .NET model and the C# language

Your sentence is missing something: the superiority of C# *compared to what*??
Or you believe that C# is the best language currently?

I agree that C# is superior to C++ for ease of development (but not for speed and memory usage),
but I don't find it superior to Scala.
A small intro to Scala: http://www.artima.com/scalazine/articles/steps.html [artima.com]

Re:

[renoX]

>C# is like every other system out there, it has its own ways and means - it can be incredibly efficient and fast, if you use it correctly.

Look, the GP said that C# is superior, this isn't even English: I asked superior in what to what?

In memory usage and performance, certainly not superior to C++ on Linux:
http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=csharp&lang2=gpp [debian.org]

And as a language, I think that Scala is better designed, that's all.
This doesn't mean that C# is bad, it's pro

Re:

[pembo13]

.NET is nice. But it doesn't seem such a massive improvement as some people make it seem. I use .NET a lot, and I have touched Java and I am familiar with C++. I don't see much that .NET exclusively offers. My favourite language is still Python however. And I may be letting my perception of Microsoft and the destruction of Delphi cloud my judgment. .NET is nice on Windows, definitely. But on Linux, there are so many more alternatives.

Re:

[mounthood]

Check out Vala [gnome.org]: C# like syntax, no runtime, FOSS.

I think...

[TaleSpinner]

> I'm thinking about developing a split personality to deal with this paradox.

I think both of you already have. =)

Re:Don't.

[0xABADC0DA]

As a developer, I am so completely convinced of the superiority of the .NET model and the C# language (Anders Hejlsberg has been my hero ever since he developed Delphi during his Borland years) that I cannot help but prefer its usage, even in a Linux environment.

If .net took away your free will it doesn't sound like you had much of it to begin with.

After reading the interviews on artima.com with Anders it's pretty clear that he's mostly a blowhard -- like Beck, all style and no substance. Take this [artima.com] for example:

Bill Venners: ... you said: "We can observe that as people write code in Java, they forget to mark their methods final. Therefore, those methods are virtual. Because they're virtual, they don't perform as well." ... Another thing that happens in the adaptive optimizing JVMs is they'll inline virtual method invocations, because a lot of times only one or two implementations are actually being used.

Anders Hejlsberg: They can never inline a virtual method invocation.

Bill Venners: My understanding is that these JVM's first check if the type of the object on which a virtual method call is about to be made is the same as the one or two they expect, and if so, they can just plow on ahead through the inlined code.

Anders Hejlsberg: Oh, yes. You can optimize for the case you saw last time and check whether it is the same as the last one, and then you just jump straight there. ...

Not only is Anders being handed his hat by the interviewer, but he doesn't even realize that for almost ten years all Java methods have been effectively final until overridden, doing exactly that which he says is impossible. There is no 'same as last time' check since it is compiled as if there was one function. There is no 'jump straight there' when the method is inlined, and no performance lost from virtual methods that are not overridden. In contrast, .NET could not even inline more than one method deep, or methods with over 32 bytecodes, or methods with any flow control whereas Java was inlining virtual methods as much as 5 calls deep.

None these guys on the core C# team were anywhere close to the same level as say Joy or Bracha, and it shows. Yeah, if you're coming from Win32 and MFC then C#/.net seems pretty awesome, but it's not. They made a lot of really bad design decisions that make it really suck compared to what it could be and what they ripped it off from. So it kind of depresses me a little bit when people gush quixotic about how great .NET is and how awesome the designers are -- it isn't, and they aren't.

Re:

[AppleOSuX]

Wow, you took a miscommunication about a very complex technical feature and turned it into "getting his hat handed to him"?

Furthermore, Anders is actually correcting the interviewer because the JVM isn't "inlining virtual method invocations", it's optimizing them.

Re:

[try_anything]

The method containing the call is speculatively compiled with the callee inlined, just like an inline method in C++, except that the inlined code is preceded by a check that the inlining is valid. If the inlining is not valid, I'm not sure exactly what happens, but the end result is that the calling method is recompiled on the fly with the callee no longer inlined.

The payoff is that many virtual method calls run just as fast as nonvirtual method calls. This is a case where Java beats C++ (or rather, the J

Re:Don't.

[Actually, I do RTFA]

For inlining a function to mean anything, it cannot be virtual. An inlined function is in essence a macro.

And Anders is more accurate than Bill. What Bill says translates to: you can turn a virtual function into an inline function with a if ( this_class__is_not_X_or_Y ) jump_to_virtual_function_table type structure. What Andres says translates to: Sure, in that case. But it's an optimization.

I see that you get the best of both worlds in that situation, but Andres is right in his unequivical "You cannot inline a virtual function." Because you cannot. You can inline exactly one version of a virtual function with an if at the top, or more if you use a bunch of ifs/a switch. But the latter removes the benefits of inlining. So, a clever compiler can have a sometimes inlined response.

And the begining of your critique is just as poor. He said that it was impossible to inline a virtual function. Stating that most functions are not virtual attacks the fact, not the logic. He said "this [disproved emperical condition] results in poor performance because of [insert logical rule] makes it impossible." Disproving the emperical condition does not disprove the logical rule.

Re:

[0xABADC0DA]

It was Anders that specifically mentioned that methods that lack final "don't perform as well" when there is no difference to the JVM until a class is loaded that overrides the method. Methods that are final could be overridden just the same, except the VM prevents it because final is about security and correctness, not performance. So from the start Anders was either blowing smoke or actually did not understand this.

I see that you get the best of both worlds in that situation, but Andres is right in his unequivical "You cannot inline a virtual function." Because you cannot. You can inline exactly one version of a virtual function with an if at the top, or more if you use a bunch of ifs/a switch. But the latter removes the benefits of inlining

Wrong on so many counts. If there is a virtual method and only one implementation, it c

Re:

[donscarletti]

However, that involves changing the code of a function (inlining it), and creating a new version for that call

Yep, that's what inlining is about, you compile a specialised version of a function as a part of its caller so that the optimising logic can see it as a single context and make speed gains across what formerly would be the boundaries between the functions. It generally wastes a little bit of memory but often gains some valuable efficiency in respect to execution time.

Avoiding a jump is something ma

Re:

[Actually, I do RTFA]

Yep, that's what inlining is about, you compile a specialised version of a function as a part of its caller so that the optimising logic can see it as a single context and make speed gains across what formerly would be the boundaries between the functions. It generally wastes a little bit of memory but often gains some valuable efficiency in respect to execution time.

Right. That means you cannot compile versions of the the program with virtual inline functions unless every permutation is included. Depend

Re:

[cyborch]

Depends. Cell processors [...]

And we all know that cell processors are widely used as a platform for CLR implementations... Oh wait, it's not. Let's stick to x86 since that's what's relevant here...

Re:

[try_anything]

What you are trying to say is that the compiler will automatically treat non-overwritten 'virtual' functions as non-virtual. I agree. This has to do with how the compiler works, not whether you can make virtual functions inline.

This involves more than just the compiler, since new classes may be loaded at any time. The JIT compiler might speculatively inline a method call only to have that method overridden by a newly loaded class. It requires runtime support to detect and correct this.

Hence, if it is virtual, you need either a) a giant block of code with switches or b) an if that jumps to a non-inlined version of the function.

Optimizing the most common case can be a big win, so option b) is not to be sniffed at. Odds are one implementation will get called most of the time. If you guess wrong, you can detect the mistake and recompile. You don't have to generate code

Re:

[Actually, I do RTFA]

I know what you're getting at here -- Java variables are all references and method parameters are all reference values allocated on the stack. However, the parent said objects, not variables, and he is correct that stack allocation is a nice optimization that cannot be done if you pass a reference to the object to a method that you know nothing about.

Wow, really? I actually still primarily use C/C++, so I'm used to explictly scoping objects, meaning that they are all declared on the stack if they are pass

Re:

[mike_sucks]

ZOMG! I can't believe you think Beck has no substance. Sure, Midnight Vultures and Guero were "like, totally dreamy" and I haven't heard the new album(s?), but everything else had more substance than a brick through a window.

Have you heard Stereopathetic Soulmanure?

/Mike

Re:

[bigstrat2003]

As yet another irrational FOSS zealot, I am philosophically opposed to Mono as an unwanted Windows fingerprint on my completely FOSS system (except for my nVidia drivers; alas, I am not perfect).

Fixed that for you. It's your machine, so obviously it's your right to run it how you want, but it also sort of marks you as a zealot who shouldn't be counted on for rational discussion of these matters.

Re:

[halivar]

Except for the part where I said a Microsoft language was the most superior available? That's too zealously anti-Microsoft for you to include me in a "rational discussion of these matters?" Holy crap, the fanboys have gone off the deep end.

Re:

[bigstrat2003]

I didn't say your zealotry was anti-Microsoft. You are striving to have only FOSS software on your system, and implied that you refuse to use closed software even if it's useful to you (unless no other option is available). That's the zealotry to which I refer, and it is quite irrational (picking software on any basis more than whether it meets your needs is nothing short of idiotic, in my opinion). Thus, it calls into question your ability to discuss other things rationally, because people aren't usually i

Re:

[halivar]

You are striving to have only FOSS software on your system

Yep.

and implied that you refuse to use closed software even if it's useful to you (unless no other option is available)

Except I explicitly stated otherwise.

Re:

[bigstrat2003]

...an unwanted Windows fingerprint on my completely FOSS system...

To me, that says zealotry, but I guess I'm quibbling. If you do not, in fact, value software license above software functionality, then I apologize. It's just an attitude that's entirely too common amongst open-source advocates here, and has really got under my skin over time.

Re:

[halivar]

License, for me, is part and parcel with functionality. It's not the only part, however (though it is large; when I say I have philosophical bias, that bias is based on what I believe are tangible capitalistic advantages of such licenses, not the crypto-anarchism often ascribed to FOSS proponents).

In those cases where proprietary technology presents a clear technical superiority to its FOSS counterpart, even despite what benefit its license might convey, I will use the proprietary technology. In my first po

One word:

[halivar]

Patents.

And yes, they exist on elements of the ECMA-334 and ECMA-335 specifications.

Re:

[TheRaven64]

I wonder how many people moderated your first post troll and flamebait as a proxy for moderating the original question as a troll. It's pretty hard to view it as anything else, when there is no real justification for actually wanting Linux given.

Re:

[Malevolyn]

I don't see how this is a flamebait. He's absolutely right. I use Linux for Linux development, and Windows for Windows development. The only exception is when I'm scripting in Python or something, then I can just check os.name or the like.

Re:

[Anonymous Coward]

I don't see how this is a flamebait.

Astroturfing is real.

Mono != .NET

[nguy]

If you want to use Linux, don't plan to develop for it in a Windows-bound way, and don't expect Linux developers to support Windows infrastructure on their system. Mono is Miguel's pet project, and no one outside Novell and Microsoft cares about it.

Mono has little to do with a "Windows infrastructure". Standard Mono installs on Linux contain the ECMA C# runtime and libraries, plus Gtk# and a lot of other FOSS libraries. .NET applications don't run on Mono out of the box.

If you want to run .NET apps, you ne

Re:

[bigstrat2003]

Bull. I, for one, care. I like .net development a great deal, and watch the mono project with a great bit of interest, because if it succeeds it'll mean that I can, if necessary, develop cross-platform apps with my preferred toolbox. Or hell, even just develop apps for my own use, with the satisfaction of knowing that any interested Linux users will also be able to enjoy the fruits of my labor.

And I'd bet money that I'm far from alone in thinking this.

Re:

[aldousd666]

I'm in the same boat with you. I'm a .Net dude for windows, but it's a huge gear shift to go to C++ or C again when I jump back to linux, so I'd like to have the 'same platform' (even if that requires quotation marks) on both OS's if I could as well.

Re:

[try_anything]

So don't go back to C++. If you aren't writing little programs that need to start up and shut down quickly, use one of the many languages available for the JVM.

For nice-looking cross-platform apps based on native GUI elements, use the Eclipse RCP (Rich Client Platform) and get a ton of other functionality for free.

Java is a bit inferior to C#, but Java and C# are just the least-common-denominator systems programming languages on their platforms. There are better languages you can use -- popular alternativ

Re:

[DeVilla]

Bull. I, for one, care. I like .net development a great deal, and watch the mono project with a great bit of interest, because if it succeeds it'll mean that I can, if necessary, develop what I believe to be cross-platform apps with my preferred toolbox. Or hell, even just develop apps for my own use, with the satisfaction of believing that any interested Linux users will also be able to enjoy the fruits of my labor.

And I'd bet money that I'm far from alone in believing this.

There. Fixed that for you.

Re:

[bigstrat2003]

That will never happen.

I'm inclined to agree, but that wasn't what you said. You said no one outside of Novell/Microsoft cares, and I was addressing that ludicrous claim.

Not that lack of applications made by .net-only programmers would be a great loss on any platform.

Nice... not only are you implying my programming skills suck (for no reason), but working under some false assumption that choice of language is in any way related to how much of a bad programmer you are! What a gem.

Re:

[bigstrat2003]

That would like saying that I care about Britney Spears because it would be nice if she was a talented, well-adjusted person with great taste.

If I had an interest in her becoming such a person, and if she were at least making strides toward becoming such a person, you really think it would be unreasonable for me to care about her efforts to reach that goal? Wow. I guess you only believe in paying any attention to goals which are a sure thing, no matter how greatly they may be of interest to you. Efficient, I suppose, but not very rewarding in my opinion.

If you are so tied to a particular language/infrastructure that you can't use anything else even if your goal requires it, you can't possibly be a good programmer

Did I say that? No, I did not. I said I like .net, and consequently, I have great interest i