05.18.08
Mono Developers: From .NET Boosting to Java Bashing?
“Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers, Developers.”
Some weeks ago we noted that Alex Brown, coming to his own defense, would attempt to bash ODF. He soon did [1, 2], which put under great shadow his role at the defunct ISO. Perhaps he was just joining Microsoft’s anti-ODF smears [1, 2] and overseas crusades [1, 2] (among other obnoxious things [1, 2]).
But in any event, this post’s focus mustn’t be document formats; it’s really about programming languages and development frameworks. The common theme here is that people justify their own choices by attempting to convince themselves that some hypothesis is correct, using flawed benchmarks such as Brown’s.
“Mono isn’t free lunch. This isn’t a free desktop.”We previously explained just how Novell helps Microsoft fight the GPL-licensed Java [1, 2, 3] and promote XAML. Miguel de Icaza last did this yesterday in his blog where he raved about Silverlight 2.0.
We also wrote about and how GNOME was getting saturated with Mono, never mind the uncertainty that's looming (yet conveniently ignored). It has already sneaked into GNU/Linux distributions other than Novell's. Remember that Mono is a Novell project, which it hopes to exploit in order to gain advantage (potentially putting others at risk).
A reader has just buzzed us to say that Mono’s more prominent promoters have just proceeded to what seems like further demotion of Java. They apparently try to show that Java is slow in order to justify their preference for Microsoft technologies.
Here is a thought: What might we be seeing here? GNU/Linux (or plainly cross-platform) developers choosing a ‘catch-up mode’ clone from a fierce and aggressive rival over an established (and original) framework that is wholly licensed under the GNU GPL? With friends like these, who needs enemies? They seem to insist strongly enough on making the Free desktop just another Windows clone with tools that are merely a compromise residing in the shadow of Microsoft lawyers. Mono isn’t free lunch. This isn’t a free desktop. █
Highlight: Novell was the first to acknowledge that Microsoft FUD tactics had substance. Novell then used anti-Linux FUD to market itself. 
Highlight: Xandros let Microsoft make patent claims and brag about (paid-for) OOXML support. 
Highlight: Linspire's CEO not only fell into Microsoft arms, but he also assisted the company's attack on GNU/Linux. 
Highlight: Microsoft craves pseudo (proprietary) standards and gets its way using proxies and influence which it buys. 
Highlight: The invasion into the open source world is intended to leave Linux companies neglected, due to financial incentives from Microsoft. 
Analysis: Xen, an open source hypervisor, possibly fell victim to Microsoft's aggressive (and stealthy) acquisition-by-proxy strategy.
AlexH said,
May 19, 2008 at 2:21 am
“As you may have seen in my second comment, I started checking some of the other tests and the Mandelbrot test for example is still a clear victor over the C# implementation. I even tried rewriting the C# Mandelbrot test to be an exact port of the Java implementation in case that made a difference, and still Java was 2x faster.”
I usually think “bashing” means “talking about the negative without acknowledging the positive”.
Roy Schestowitz said,
May 19, 2008 at 3:58 am
Alex Brown too criticised MSO07 (not OOXML mind you) just before bashing ODF. Let’s wait and see.
AlexH said,
May 19, 2008 at 4:09 am
But the thing is, the criticisms he made of ODF weren’t wrong: he’s entirely right that basically no application outputs ISO standard ODF.
It’s entirely enlightening that you class negative remarks about OOXML as “criticism”, but when they’re about ODF they’re “bashing”.
Roy Schestowitz said,
May 19, 2008 at 5:07 am
No, that’s the disinformation. See Rob Weir’s response to this.
AlexH said,
May 19, 2008 at 5:32 am
Rob’s response is inadequate because he uses the oasis files, not the ISO ones.
Leaving aside the purely technical aspects of whether or not there are issues with the ISO ODF standard, which is Alex Brown’s point, the inescapable fact is that apps like OpenOffice.org are using OASIS ODF 1.1 (which isn’t ISO standardised), and 3.0 is using 1.2 which isn’t even OASIS standardised yet.
They all output stuff which isn’t in the ISO standard, although it looks like OOo 3.0 might be gaining an option to output ISO standard files.
Does it matter? Not really. Is Alex Brown wrong? No.
Dan O'Brian said,
May 19, 2008 at 7:34 am
Roy: Do you have some sort of personal vendetta against the Mono developers or something?
It seriously looks like you are desperate to prove how evil they are, you didn’t even read the article.
Xanadu said,
May 19, 2008 at 7:48 am
I don’t really get why would anyone jump to defend OOXML unless there were lame intentions behind. I specially would be skeptic if the same person that protects Mono because it is open source software made by innocent developers would also defend OOXML while attacking or echoing ODF attacks.
It is sometimes too much of a coincidence, a person promoting Mono at the same time attacking ODF, defending OOXML and finally advocating Novell, and their deal. It looks like all those things come in the same package.
AlexH said,
May 19, 2008 at 8:10 am
@Xanadu:
It’s just simply not so black and white. For example, Alex Brown criticised the ODF schema in its ISO 1.0 incarnation. What then did he do? He published a revision to the schema which he believes removes/fixes the problem.
That’s constructive criticism that helps ODF grow stronger. Calling it ‘disinformation’ or ‘ODF bashing’ just highlights bias and misunderstanding.
Roy Schestowitz said,
May 19, 2008 at 8:38 am
Alex, how much do you know about Brown’s business? Have you followed the links? The man needs to justify poor recommendations that he made. He told the British Library to shove its huge assets right into vendor lock-in. Many people are absolutely furious over this, and rightly so.
Roy Bixler said,
May 19, 2008 at 9:17 am
Brown did make the rather bold claim that “there are no valid ISO ODF documents in the world.” Rob Weir disproved that claim by coming up with a minimal valid ISO ODF (v. 1.0) document. Even if Weir’s post didn’t clear up all of the issues, such as that OpenOffice uses the Oasis version of ODF, his post did provide some needed clarification to Brown’s claims.
AlexH said,
May 19, 2008 at 10:12 am
@Roy B: indeed, that claim was wrong. He would have been better to say ‘no valid non-trivial document’ - it’s basically impossible to avoid the schema problem in those instances.
@Roy S: How much do I know about his business? How much do you know?
“Don’t believe everything you read on the Web, Karsten. As it happens I am not contracted to the British Library (there are some lies in circulation to the contrary for consumption by the credulous). Wish I was though - I believe they have a lot to gain from the expertise and technology my company offers!”
– Alex Brown, http://adjb.net/comments.php?y=08&m=04&entry=entry080409-221633
So what recommendations, exactly, did he make to the British Library? What, exactly, does he need to justify?
Maximus said,
May 19, 2008 at 10:29 am
AlexH: So what’s this developer’s excuse for claiming that C# is faster than C?
There’s no excuse for presenting falsified data like this.
AlexH said,
May 19, 2008 at 10:34 am
@Maximus:
So, you did that test yourself and you know the data is falsified?
I really don’t think so.
I mean, good grief, it’s well known that in certain circumstances dynamically compiled languages can outperform statically compiled languages. It’s not just Mono/.net; Java is much faster than C in many well-known situations, see e.g. http://www.idiom.com/~zilla/Computer/javaCbenchmark.html
Maximus said,
May 19, 2008 at 10:44 am
I don’t trust anything those Mono guys do. They are known to spread lies and work against the Free Software community. It’s all well documented.
I would test the results but I don’t have Mono installed and I will never install it.
Roy Schestowitz said,
May 19, 2008 at 10:46 am
Benchmarks are easy to tweak. Just ask analysts… or even Microsoft. They can show anything and I doubt Jeff’s hypothesis is that Java beats Mono.
In fact, Mono’s bad reputation is that it’s slow and heavy, so even if he shows parity, then he markets Mono.
AlexH said,
May 19, 2008 at 10:47 am
So, in short, you’re calling him a liar even though you have no idea whether or not the data presented is correct. Nice.
If I repeat the test would you believe me?
Roy Schestowitz said,
May 19, 2008 at 10:49 am
Alex, you’re biased too, IMHO, so replicating results won’t do.
AlexH said,
May 19, 2008 at 10:57 am
Roy:
That’s a convenient excuse to ignore actual, repeatable, data.
And actually, I imagine Jeff’s hypothesis was that Java was faster than Mono. He said specifically he was looking for the bug in Mono that made it slower than Java - it just turned out that the data disproved his hypothesis.
Of course, benchmarks are pretty meaningless. However, these weren’t the Mono project’s benchmarks - they were someone else’s. So, what you’re complaining about is actually “Mono project members improve their software” - how terrible of them!
Miles said,
May 19, 2008 at 11:06 am
And here we see the crux of the issue.
The BoycottNovell bigots are always quick to accuse the other side of being liars, but are never willing to review the other sides evidence.
Afraid you’ll get proven wrong?
How typical.
Miles said,
May 19, 2008 at 11:12 am
FWIW, I just ran the programs on my machine and got the following results (looks like my machine is slower than his, but the results more or less match up):
[miles@localhost ~]$ time mono sumcol2.exe
Miles said,
May 19, 2008 at 11:13 am
FWIW, I just ran the programs on my machine and got the following results (looks like my machine is slower than his, but the results more or less match up):
[miles@localhost ~]$ time mono sumcol2.exe < sumcol-input100000.txt
50000000
real 0m3.799s
user 0m3.476s
sys 0m0.316s
[miles@localhost ~]$ time ./sumcol < sumcol-input100000.txt
50000000
real 0m18.303s
user 0m18.001s
sys 0m0.284s
(sorry for the repost, had to html encode the less-than char)
Dan O'Brian said,
May 19, 2008 at 3:39 pm
hehe, I love the comment on that guy’s blog that quotes Princess Bride. Looks like it holds pretty true, too
I might have to add his blog to my rss feed.
Dan O'Brian said,
May 19, 2008 at 5:58 pm
So Roy and Maximus - you gonna admit you were wrong? Or are you gonna provide some evidence that this developer is falsifying data?
I’ll be waiting.
Roy Schestowitz said,
May 19, 2008 at 6:02 pm
Miles,
All these things are implementation/program-dependent. One could prove almost everything that’s desirable, so benchmarks are a more complex things than that. I’m a technical researcher and I know that peer review would scrutinize for exactly this reason. One quote that also comes to mind:
“Microsoft did sponsor the benchmark testing and the NT server was better tuned than the Linux one.”
http://www.itweb.co.za/sections/enterprise/1999/9904221410.asp
Dan O'Brian said,
May 19, 2008 at 6:06 pm
This benchmark test wasn’t designed by the Mono guys.
You are just trying to find excuses now, pretty pathetic.
Roy Schestowitz said,
May 19, 2008 at 6:07 pm
Who ever claimed falsification of data?
Dan O'Brian said,
May 19, 2008 at 6:29 pm
Maximus claimed falsified data and you claimed the guy’s results couldn’t be trusted (which suggests the same thing).
You refused to run the tests yourself and so AlexH offered, to which you claimed he was biased and so results from his reproduction couldn’t be trusted either.
Are you now saying that you trust that the results are indeed accurate and that the guy’s C# implementation was 6-7x faster than the fastest C implementation on the Debian Language Shootout site?
Dan O'Brian said,
May 19, 2008 at 6:39 pm
FWIW, the developer’s blog you quoted never once draws the conclusion that this test conclusively proves that Mono is faster than Java or C, he simply concludes the following:
1. that the brain-dead ReadLine() + Int32.Parse() implementation of THIS PARTICULAR TEST in C#/Mono can run as fast as the same brain-dead ReadLine() + Integer.Parse() implementation in Java. (as in, comparing apples to apples)
2. that he can write a much more optimized C# implementation that can outperform the fgets_unlocked() + atoi() implementation in C (which he successfully proved he could).
You also misrepresent the facts when you claim that he was bashing Java. Nowhere in his blog post did he bash Java. Nowhere.
You are scum for suggesting otherwise.
Masato Naru said,
May 20, 2008 at 5:20 am
I don’t know what that Java-guy smoked before he performed his tests, but the results are false.
Read here for: an failed attempt at reproducing his results, courtesy of Jeffrey Steadfast.: http://jeffreystedfast.blogspot.com/2008/05/debian-language-benchmarks-sumfile.html
Before accusing Mr. Steadfast of an outright lie (which I espect of you anti-Mono extremists): The data is there, the source is there; compile it and reproduce before you make any such claims!
Roy Schestowitz said,
May 20, 2008 at 7:23 am
What’s his hypothesis? Remember: we’re not talking about performance here. It’s a distraction, a decoy. See:
http://www.microsoft.com/interop/msnovellcollab/moonlight.mspx
Care to explain what this means if one didn’t pay Novell (Microsoft’s software patent royalties)?
Dan O'Brian said,
May 20, 2008 at 9:05 am
What does Moonlight have to do with the tests?
Nice try at changing the subject, Roy, but it was an epic fail.
Even if Mono was the antichrist, it still wouldn’t change the results of the tests.
Roy Schestowitz said,
May 20, 2008 at 9:10 am
The two are joined by the hip, to an extent licence-wise too. The eventual goal is to deliver applications, potentially over the NET and WPF seems to be inspiring the GNOME desktop these days.
Dan O'Brian said,
May 20, 2008 at 10:48 am
Once again you failed to provide the connection between how the results of the tests he performed are affected in any way shape or form by Moonlight (Moonlight is not Mono and licenses do not change performance results… just in case you didn’t realize that).
Miles said,
May 20, 2008 at 1:24 pm
/bump - any updates on this?
(actually, I’m just bumping to further humiliate Roy, but I can pretend that I’m actually interested, can’t I?)
Roy Schestowitz said,
May 20, 2008 at 8:46 pm
Dan, the point to be made is that the issues we should deal with aren’t to do with performance. They never were. We never said that Mono being slow is the issue, but marketing it (with the legal implications) at the expense of other PLs seems foolish and dangerous.
Sampa Mutoku said,
May 21, 2008 at 6:13 am
Don’t try to wiggle yourself out of it. You arre just traying to bash Mono where ever your find it, using whatever means you think handy.
This time you’ve chosen the ’speed/performance’-comparison as a tool, and you have failed miserably.
Note: comment has been flagged for arriving from a possible incarnation of a known (eet), pseudonymous, forever-nymshifting, abusive Internet troll that posts from open proxies and relays around the world.
Roy Schestowitz said,
May 21, 2008 at 6:17 am
Hi, ‘eet’. Nice nymshift. Your IP gives it all away, never mind the obvious giveaways in your messages.
Dan O'Brian said,
May 21, 2008 at 9:26 am
I have to agree with the above poster - you are using this as an excuse to bash Mono while trying to play it off like the Mono guys are bashing Java.
They aren’t the ones bashing anyone, you are bashing them.
JK said,
May 30, 2008 at 3:29 pm
His numbers are inconsistent by what is reported by a third-neutral party. Mono is slower than Java, at least for now, by a wide margin, according to the ranking http://tinyurl.com/lf82f